Chapter 3: Choosing a government-lower houses as electoral colleges

Prolonged post-election turmoil, with minority governments or unstable coalitions, does not sit well with the public in any of our four countries. Nor do voters appreciate too-frequent elections to resolve political instability. Voters want a stable government with the power to implement the broad policies on which it has been elected. This does not mean that voters necessarily support all the items in a government’s election policy-the so-called ‘mandate’ much beloved by politicians-but they do want a clear election result and a government which is preferred by a majority of voters, and we should look first at how successful the various electoral systems have been at achieving this.

The first issue is whether the country has been fairly divided into equal electorates, and also whether the voting system produces the desired result. The question of public financial support for election campaigns should also be considered, to ensure that in these days of mass media, victory is not almost automatically to the richest party.

Of course elections will not always be decisive, and sometimes MPs prefer minority governments to another election. Some of these minority governments have been surprisingly successful.

The parliament of course retains the power to remove a government by passing a vote of no confidence in it. Parliament has less say in a change of prime minister or premier, with the decision being left to the government party.

The twenty parliaments we are considering have various terms, varying from three years to five. It is worth looking at the various terms to see which is best. Then there is the question of the desirability of a fixed term for the parliament, removing the power of the prime minister or premier to cut short the term for political advantage.

Then there is the possible role of the upper house in forcing an unwanted election on the lower house. How and when (where it exists) could this power be used?

Finally, what is the role of the head of state in appointing or dismissing the prime minister or premier, and also in approving the dissolution of parliament?

The US model

In the United States the role of choosing the president (who is both head of government and head of state) is not normally performed by the legislature, the Congress, but by a separately-elected body called the Electoral College. In the Westminster system, the electoral college role in the choice of head of government is performed by the lower house of the legislature.

The original concept of the American Electoral College was that delegates from each state should meet, discuss and choose the most suitable individual to be president, but inevitably candidates soon began standing for the Electoral College pledged to vote for a particular individual. In most of the states there is not even the pretence of voting otherwise. There are 538 electors, each state having a strength equal to the number of its senators and congressmen, and the District of Columbia, which has no voting representation in the Congress, has three Electoral College votes. The election in each state is on a ‘winner takes all’ basis, that is, the presidential candidate who gets the most votes in a state is deemed to have all that state’s Electoral College votes. There is no central authority conducting the elections. There are 50 quite independent state electoral authorities, all operating under different laws and supervising various methods of casting and counting votes, which concern not only the presidential elections but also elections for Congress, state governors and legislatures, and local legislatures. Most of the states have passed the responsibility for the voting systems down to the counties and cities, some 3000 of them, who have to pay for the voting methods they choose.

A presidential candidate needs 270 Electoral College votes to be elected. If no presidential candidate has that many, and cannot gain them by negotiation, the election is decided by the House of Representatives from among the top three candidates, each state having one vote. Because of the way in which the Electoral College delegates are chosen, it is possible for someone to gain the presidency with fewer nationwide votes than another candidate. This occurred in the 2000 presidential election, when George W. Bush won a majority of votes in the Electoral College after prolonged legal disputes, but was 500 000 behind the other major candidate, Al Gore, in the nationwide vote. This has happened on only three other occasions-John Quincy Adams (1824), Rutherford B. Hayes (1876) and Benjamin Harrison (1888). The use of the House of Representatives to resolve a deadlock in the Electoral College has been necessary three times, in 1800, 1824 and 1876. Occasionally there have also been electors who have broken their voting pledge (this has happened six times in the past ten presidential elections) but it has never affected the result.

One further point should be made. Because neither voting nor even registration as a voter is compulsory (and the latter is sometimes administratively tedious), voter turnout is rarely more than 50 per cent. The 64 per cent turnout in 1960 was the highest in modern times.

So much for the American system. Turning to the Westminster system, Bagehot thought that choosing the executive was the most important task of the lower house of parliament. Let us look then at how well the various lower houses have performed this task between 1970 and 2000.

A fair and decisive result?

Most voters in the four countries we are considering seem to want the result of an election to be a decisive choice of government, and it is obviously desirable that the government chosen should, if possible, be one preferred by a majority of voters. It is also important that the voters should have sufficient information, particularly in the financial area, to enable them to judge the performance of the incumbent government and its future plans. Governments often suppress such information, and sometimes issue grossly misleading forecasts.

United Kingdom

There were eight elections between 1970 and 2000. Seven of them were won by a party with a workable majority in the House of Commons, together with the largest percentage of the national vote. There was one minority government after an election. In February 1974, Labour won 301 seats out of 635, the Conservatives 297, the Liberals fourteen, Ulster Unionists eleven, Welsh and Scottish Nationalists and others three. Labour was in fact 0.8 per cent behind the Conservatives in the national vote. The Wilson Labour Government was not immediately challenged in the Parliament, because no party wanted another election so soon. Wilson held on for eight months before he obtained a dissolution in October 1974. He managed to win an absolute majority of only three, despite being 3.4 per cent ahead of the Conservatives in the national vote.

There were no coalition governments during the three decades, nor were any seriously contemplated as a means of dealing with a minority government or one with an unworkably small majority. It is true that Conservative Prime Minister Edward Heath (1970-74) negotiated with the Liberals before resigning after losing the February 1974 election, and there was a Liberal-Labour pact to support the October 1974 Labour government after it lost its narrow majority. This pact lasted from March 1977 until May 1978. The Liberals gained no ministerial offices from the pact, but perhaps benefited slightly from regular consultations with ministers. In any case neither Labour nor the Liberals wanted an election.

It is worth noting that none of the eight governments won more than 50 per cent of the national vote. The highest was 46.4 per cent by the Conservatives under Heath in 1970 and the lowest 37.1 by Labour under Wilson in February 1974. The reason is of course the presence of the Liberals and the various national parties. The Social and Liberal Democrats are the result of the merging of the Social Democratic Party (a right wing breakaway from the Labour Party) and the Liberals. The nationalist parties are the Plaid Cymru (in Wales), the Scottish National Party and the Ulster Unionists. The Ulster Unionists have close links with the Conservative Party.

The voting peak for the Liberals and Social Democrats was 25.4 per cent of the vote in 1983 (which gave them 23 seats) and for the nationalist parties 6.7 per cent in October 1974 (which gave them 26 seats). The Liberals always suffered in representation because their support, though sometimes substantial, was always diffused. Moreover the percentage of votes won by the minor parties is misleadingly low, because they do not contest all seats. The major parties, on the other hand, are becoming more geographically concentrated, with Labour the party of Scotland, Wales and Northern England, while the main strength of the Conservatives is in the south.

It cannot be said that the House of Commons performs very well as an electoral college making a decisive choice of government which reflects the wishes of those interested enough to vote. After all, since 1970 there has been one elected minority government and one with such a narrow majority that it soon became a minority. On the credit side, voter involvement is much higher than in the United States, where only about half those eligible actually vote. In the UK the typical figure is three-quarters, ranging from a post-war high of 84 per cent in 1950 to a low of 71.5 per cent in 1997. In the 2001 election the figure fell even further, to 59 per cent, the lowest since 1918. Undoubtedly the overwhelming support for the Blair Government was the key factor in voter apathy, but it must be a worry when an MP can be elected on a voter turnout of 39 per cent, as happened to a Labour MP in Shropshire, and when nearly eighteen million of Britain’s 44 million registered voters abstained. Fewer than one in four eligible voters under 25 cast a ballot.


Although the Canadian House of Commons is elected for a five year term, there were seventeen elections between 1945 and 2000, so the average life was little more than three years. Of these seventeen elections four were inconclusive, with no party having an absolute majority. When this arises the custom has been followed that if the incumbent government is the largest party after the election it remains in office and faces the House of Commons, while if the opposition is the largest party, the government resigns and the leader of the opposition is invited to form a government.

The parties in the House of Commons are numerous, and tend to be geographically based. The Liberal Party’s political philosophy has been middle-of-the-road, trying to project an image of competence, and to appeal to the middle class of Ontario and Quebec, and to francophones. The Progressive Conservatives are basically a conservative party, with the progressive part of the name the result of an earlier amalgamation. Their power base used to be Ontario and the prairies, but they were almost annihilated in the 1993 election and are recovering slowly. The Reform Party (now the Alliance Party) was established in 1987. It is a right wing populist party, anti-French and anti-Ottawa. Its power base is the western provinces, and it is taking many votes from the Progressive Conservatives. There have been moves to unite with the Conservatives, but so far these have come to nothing. The Social Credit Party owes little to the fundamentalist economic policies of Major Douglas, and is now conservative both socially and economically. The New Democrats (CCF/NDP) Party is the equivalent of the Labour parties in the other countries, though perhaps rather more left wing. The Bloc Qubcois represents the interests of Quebec only, and was sufficiently strong to become the official opposition for a time.

An inconclusive election obviously offers a wonderful opportunity for deals between the major and minor parties, but Canada has developed some unique ground rules. There is never a suggestion of a coalition between the major parties. There is never a suggestion of a minor party going into coalition with a major party to form a government. Instead, the minor parties look to influence the government program so as to achieve some of their political objectives. Finally, the minor parties and Independents are aware that in an election to resolve a hung parliament the voters tend to look to the major parties, and the minor parties suffer. This given them a strong incentive not to force a premature election.

The Canadian government has an unusual discretion in deciding the date of a by-election to fill a casual vacancy in the House of Commons. A seat may remain vacant for many months if public opinion is running against the government. Prime Minister Trudeau once called fifteen by-elections on the same day.

In the 1957 election the long Liberal reign came to an end. From then until 1968 there was much instability, with no less than four more elections. Some stability returned in 1968. In that year Liberal Pierre Trudeau succeeded Lester Pearson as prime minister, and called an early election. The Liberals gained 155 seats to the Progressive Conservatives 72, the New Democrats 22 and the Quebec version of Social Credit fourteen. Such stability could not last, and the 1972 election saw the Liberals and the Progressive Conservatives again neck-and-neck, 109 seats to 107, with the New Democrats holding the balance of power with 31 seats. It was touch and go which of the major parties would have the greater number of seats, one riding (the Canadian description of an electorate) being won, after recounts, by only four votes. The Liberal government faced the House and was sustained for sixteen months, though losing eight out of 81 votes, before being defeated on a vote of confidence. At the ensuing election (1974) the Liberals won an absolute majority.

Again it could not last, and the 1979 election returned to the indecisive pattern. The House of Commons was enlarged to 282 members, but the largest party, the Progressive Conservatives, won only 136 seats, and 35.9 per cent of the popular vote. Either the New Democrats (26 seats) or Social Credit (six seats) could have given the Progressive Conservatives a majority-only just in the latter case-but both would have been needed to sustain a Liberal government. In the event, the Liberal government resigned immediately after the election, and a Progressive Conservative government tottered on for a few months. The prime minister, Joe Clark, did not face the House for five months after the election, and two months later was defeated on a vote of confidence.

The Liberals won an absolute majority at the 1980 election, and set the pattern for the next four elections-Progressive Conservatives in 1984 and 1988, and Liberals in 1993 and 1997, though only the Progressive Conservatives, with 50 per cent in 1984, won a majority of the popular vote.

The 1993 election was dramatic, with the governing Progressive Conservatives being reduced from 151 seats to two. It also marked the defeat of Mrs Kim Campbell, Canada’s first woman prime minister. She had held office for only a few months, after the resignation of her predecessor, Brian Mulroney. The Liberals gained a massive majority, gaining 177 seats in the 295 seat House of Commons. The Liberals swept Ontario and Atlantic Canada, winning all but one seat in each of them, and won an unprecedented number of seats in the west. The election marked the rise of two new regional parties, the Bloc Qubcois with 54 seats, and the Reform Party with 51 seats, all but one in the west and mostly taken from the Progressive Conservatives. The Bloc Qubcois, as the largest non-government party, became the official opposition. The party leader gave an undertaking that the Bloc would not be obstructive in the House just to show that Quebec would be better off as a sovereign state because federalism did not work. Neither the Progressive Conservatives nor the New Democratic Party achieved the twelve seats required for recognition as official parties in the House.

The general election in 1997 saw the Liberal Party win a second consecutive majority government, but the Liberals were reduced from 177 to 155 seats, giving them only a narrow majority in the 301 seat House of Commons. The Liberals’ strength came largely from Ontario, and many seats were lost in Atlantic Canada, where the seats held fell from 31 to eleven out of 32 available. Two Cabinet ministers lost their seats, and all the Liberal seats in Nova Scotia were lost. The Reform Party increased its numbers to 60, all from the west, and became the official opposition. The Bloc Qubcois fell from 54 to 44 seats, and from 49 to 38 per cent of the vote in Quebec, which meant that they won less than half of the votes of the francophones. The Progressive Conservatives staged something of a recovery from their 1993 disaster, and won 19 per cent of the popular vote and twenty seats, chiefly in Atlantic Canada. The conservative vote in the west mostly went to the Reform Party.[24] The New Democratic Party also staged a comeback, increasing its numbers from nine to twenty.

Canada has made a decisive choice of government in each election of the last two decades of the century, but the lack of effective nation wide parties competing for government is damaging to national cohesion.


All the 22 national elections in Australia between 1946 and 2000 were fought out between the Labor Party and the Liberal-National Party Coalition, with the exception of the 1987 election when the Coalition split, and lost an election it should have won. The Liberal Party is roughly equivalent to the British Conservative Party, and the Labor Party (the ALP) to the British Labour Party. The National Party (formerly the Country Party) is a very conservative rural party, in favour of considerable government involvement in rural marketing. There have been occasional Independents and minor party MPs, but since 1943 they have never held the balance of power in the House of Representatives. Each election has resulted in an absolute majority, and none of the governments has had any real difficulty in getting its legislation through the House, even when there was only a narrow majority, as there was in 1961-63, when the coalition government had only a one seat majority. Such is the power of party discipline.

Unlike Canada, the major parties are nation wide. The balance between Labor and the combined Liberal and National Party votes is remarkably consistent around Australia. But although decisive electoral results are expected, it is not the case that the combination of single member electorates and preferential voting has always produced the government desired by a majority of voters. In 1954, 1961, 1969 and 1998, when the Coalition won the elections, a narrow overall majority of voters would have preferred a Labor government, and in 1990 the Coalition was just preferred by the voters, but Labor won government with a safe eight seat majority. Single-member electorates disadvantage parties such as the Australian Labor Party, which tends to have excessive numbers of voters (many more than are needed to win) concentrated in electorates in industrial areas. Parties such as the Australian Democrats, with a relatively small number of widely scattered voters, are also disadvantaged.

An interesting new party emerged in 1996. Pauline Hanson had been selected as the Liberal candidate for the safe Labor seat of Oxley, in Queensland. During the campaign she wrote an anti-Aboriginal letter to the local newspaper, and she was disendorsed by the Liberal Party. She continued as an Independent, and won the election, which she never would have done as a Liberal. The reason for her success was that she advanced ideas which were firmly held by many country people, but which were unacceptable both to Labor and the Coalition. She campaigned against more Asian immigration, and for fewer benefits for Aborigines, cheaper loans for struggling farmers, and no tightening of the gun ownership laws. Her position as an Independent in the House of Representatives was used to form the One Nation Party, but as the party developed and began to produce broader policies-on taxation, for example-the intellectual limitations began to appear. Nevertheless One Nation managed to win eleven seats and gained 23 per cent of the vote in the Queensland state election in 1998, and in the federal election later in the same year, although Mrs Hanson was not re-elected and her party did not win any seats in the House of Representatives, they did manage to elect a senator from Queensland, taking a seat from the National Party. Across the country, One Nation won 8.4 per cent of the vote, considerably more than the Nationals (5.3 per cent) and the Democrats (5.1 per cent). It was typical, though, of the highly centralised and generally incompetent management of the party that the successful Senate candidate turned out to be ineligible, because she was not an Australian citizen. Her place was taken by the second candidate on the One Nation list.

Problems soon emerged in Queensland. One of the One Nation MPs resigned only four months after he was elected, and the state Labor government won the by-election, turning it into a majority government. Soon afterwards five other One Nation MPs left the party and sat as Independents, leaving only five One Nation MPs. Then the courts ruled that One Nation had not been properly registered for the 1998 state election, and had to refund the electoral funds it had been given, driving the five surviving One Nation MPs to leave One Nation and form their own party.

But the issues which led to the sensational rise of the One Nation Party have not disappeared, and it may well be a significant influence in future elections.

The federal government has taken action to ensure that the voters are properly informed on fiscal matters before an election. After winning the 1996 election it issued a ‘Charter of Budget Honesty’, to ensure ‘that no other government in the future behaves in the way the Labor government behaved in the March 1996 election’, when it ‘assured the Australian public that the accounts were in balance when in fact, as the outcome for that year showed, the government deficit was $10 billion.’

To achieve this goal, the secretaries of the treasury and finance departments are required to prepare a pre-election report providing assessments of the fiscal and economic outlook. This report is to be released within ten days after the announcement of an election. Such a report would certainly prevent the government from making unrealistic promises, and could help the opposition in developing its plans.

New Zealand

All the elections between 1935 and 1993 resulted in a clear win for either the Nationals or Labour, though the 1993 election was a very close-run thing, with the Nationals finishing with a one seat majority. Otherwise the only two elections when the party strengths were close were 1957 and 1981. In 1957 Labour had a majority of two over the Nationals, and in 1981 the Nationals had a majority of two over the combined numbers of Labour and Social Credit. Labour Party discipline in New Zealand is such that no real difficulty was experienced by the 1957 government, and the Parliament ran its usual term. National Party discipline is slightly, but only slightly, more relaxed. In 1983 two National MPs voted against key clauses of an industrial relations bill, causing a government defeat on the issue. Prime Minister Muldoon called for a pledge from party members not to cross the floor in 1984, an election year. One MP qualified the pledge by saying she would not accept the party policy on disarmament, visits by nuclear ships, and rape. Muldoon immediately called a snap election, which he lost.

The single member electoral districts and first-past-the-post voting did not always award power to the party preferred by the majority of the voters. Of the seventeen elections between 1946 and 1993, nine were clearly won by the more popular party. In the other eight elections neither of the two major parties had an absolute majority of the votes cast, minor parties receiving a substantial proportion of the votes, such as 20.6 per cent for Social Credit (and two seats) in 1981 and 12.2 per cent (but no seats) for the New Zealand Party in 1984. In the 1978 and 1981 elections the Nationals won government with an absolute majority of seats, despite having smaller shares of the overall vote than Labour.

The National Party is conservative but more rurally influenced than the Australian Liberal Party. The Labour Party is very similar to the Australian Labor Party. The Social Credit Party’s main strength was among the poorer farmers of the North Island; in all except financial matters (where it was populist) the party was ultra conservative. It was renamed the Democrats Party in an attempt to broaden its appeal, and in 1991 formed the Alliance Party by merging with the Greens, New Labour and the Maori Party. The New Zealand First Party was formed by Winston Peters after he was dismissed from the National ministry in 1991, but the party held the balance of power after the 1996 election, and Peters was re-admitted to the Cabinet only to be dismissed again. The party split, but still survives. The ACT Party is a party to the right of centre.

Concern over the lack of representation of significant minor parties led to the adoption by referendum in 1993 of the West German Additional Member System, which the New Zealanders called the Mixed Member Proportional System, usually shortened to MMP. MMP was first used in the 1996 election. The size of the House of Representatives was increased to 120 members, 65 members from individual electorates (six of them Maori), and 55 from party lists. Under the MMP system, electors had two votes. One was by first-past-the-post voting to choose a member for their particular electorate. The other vote was for a party list. Party list votes are calculated nationally and parties are allocated seats from their lists so as to make their total number of MPs equal to their percentage of the national party list vote. Parties which win less than five per cent of the national party list vote are not allocated any seats.

As was inevitable, neither of the major parties won an overall majority in the 1996 election, and the smaller parties won substantial numbers of seats. There were no less than nine parties represented in the Parliament, as well as four Independents. The National Party had 44 of the 120 seats and the Labour Party 37. The largest of the smaller parties was the New Zealand First Party, and other significant ones were the left leaning Alliance Party and the right wing ACT. After eight weeks of negotiations the ruling National Party managed to achieve an alliance with the New Zealand First Party, which gave them a majority. This was despite the fact that New Zealand First had been bitterly opposed to the Nationals during the election campaign.

Such as alliance could not last. In August 1998 the leader of the New Zealand First Party, Winston Peters, at that time Deputy Prime Minister and Treasurer, led his party members in a walk-out from a Cabinet meeting called to decide on the terms of the sale of the government shares in Wellington International Airport. The outcome was the dismissal of Peters from the Cabinet, and the dissolution of the coalition. Although four of Winston Peters’ former colleagues rejoined the government as Independents it still had only 48 of the 120 seats, relying on sufficient votes from minor parties and Independents to enable it to survive on votes of confidence.

New Zealand went to the polls for the second time under the MMP system on 27 November 1999. On election night the Labour Party and its ally the Alliance Party seemed to have won, and a formal coalition agreement was signed on 6 December and a Cabinet of twenty ministers (sixteen Labour and four Alliance) was announced. (A great improvement on the nine weeks it took to form a government after the 1996 election.) There were some delays in finalising the vote counting, because of recounts, some irregularities in procedure, and the additional work of vote counting for two citizens-initiated referendums. The Green Party’s overall vote eventually rose above 5 per cent, entitling it to have additional seats to bring its strength up to its share of the national vote. This gave the Greens seven seats, and reduced the Coalition to 59 seats in the 120 seat House of Representatives. Although the Coalition was technically a minority government, it could rely on support from the Greens on every important issue.


It can be seen that the election results for the four national parliaments have been erratic. They have by no means always produced a decisive result, which causes either a minority government or a post-election coalition. Even when they have produced a decisive result, it has not always been a decision which reflected the wishes of a majority of the voters. There are two factors which must be considered here. Are the electorate boundaries fairly drawn, without gerrymandering or malapportionment? And is the voting system employed one which is most likely to produce a fair and decisive result? Let us look at these problems in turn.

Distribution of electorates


Eighteen of the twenty lower houses we are considering now use single member electorates as the source of their MPs. The odd ones out are Tasmania, which has used proportional representation since 1909, now with five members being elected from each of five constituencies; and New Zealand, which since 1996 has used the MMP system, a mixture of single member constituencies and proportional representation. Whichever method is used, the fairness of the drawing of constituency boundaries is important.

It is claimed that in 1812 Governor Gerry of Massachusetts, a Republican-Democrat, endeavoured to have the state electoral boundaries redrawn so as to concentrate the opposition federalist vote in a few districts. One of the electoral districts was so oddly shaped that it was claimed that it looked like a salamander, and the process was named ‘gerrymander’ after the Governor. It worked. The Republican-Democrats won the election in terms of seats, in a landslide 29-11, although the federalists won more votes. It later emerged that Gerry was actually opposed to the re-distribution.

Gerrymanders survive. The US Supreme Court recently ruled that a famous Z-shaped congressional district in North Carolina was not unconstitutional because it had been gerrymandered for political rather than racial reasons.

Gerrymanders are not the only way to distort election results. Another way is to have different numbers of voters in different electorates. There is always pressure from rural electorates to be given a smaller quota of voters, partly because of the vast size of some rural electorates, and partly because country people regard themselves as the true creators of wealth. The process of varying the number of voters in electorates to meet political pressures or gain advantages is called malapportionment by the Americans.

United Kingdom

It cannot be said that the British system of drawing constituency boundaries works very well. There are four boundary commissions, one each for England, Scotland, Wales and Northern Ireland. Each has a judge as de facto chairman, two other members, usually barristers, and is assisted by the registrar-general and the surveyor-general. A commission’s first task is to recommend the total number of seats for the country for which it is responsible. This apportionment has continued to be badly skewed. The 1986 Parliamentary Constituencies Act laid down that the number of constituencies in Scotland should not be less than 71 and in Wales not less than 35, while in Northern Ireland the number should not be greater than eighteen or less than sixteen. If one allowed a uniform electoral quota throughout the United Kingdom, thirteen Scottish seats and six Welsh ones would be abolished. Alternatively, it would be possible to avoid any decrease in the number of seats for Scotland and Wales by increasing the size of the House of Commons to 783, with 129 new seats going to England, a solution which seems very unlikely.

When the Scottish Parliament was established in 1998 the Parliamentary Constituencies Act was amended to remove the guarantee of 71 seats for Scotland at Westminster, though nothing was done about the Welsh or Northern Irish quotas. The amendment also provided that the English quota should apply in deciding the number of seats for Scotland, though the Commission is required to take into account the boundaries of local government areas and geographical considerations (the amendment to the Act specifically directed that Orkney and Shetland should remain a separate constituency). It seems certain that the next review will result in fewer seats for Scotland, though the Scots will undoubtedly fight hard against any serious reduction. In any case, the Commission will not report until some time between 2003 and 2007.

The commissioners have regarded the avoidance of crossing county and London borough boundaries, and geographical considerations, as being higher priorities than achieving an equal number of voters in each constituency. The result has been that the constituency with the highest number of voters (the Isle of Wight, with 101 680 electors in the 1997 election) has nearly five times the number of the constituency with the fewest (22 938, in the Western Isles of Scotland). Moreover, redistributions are infrequent-eight to twelve years after the presentation of the previous report-and have to be approved by Parliament, with much opportunity for political manoeuvring and delay. There may be a great deal of voter movement between a commission beginning its investigation and Parliament approving the result.


Canada has had great problems in producing a fair system of distributing parliamentary seats. There must be some limit on the number of MPs in relation to population, but on the other hand provinces with a falling proportion of the national population do not like to have their number of MPs reduced, and the smaller provinces have successfully fought to retain a minimum number of MPs.

The number of federal seats for each province is determined by a complicated formula based on the decennial census population (excluding the population of the three territories) to establish the average population per seat (about 90 000). The appropriate number is then assigned to each province. The three territories each have one MP. The system is complicated by the constitutional requirement that no province shall have fewer MPs than it has senators, and the statutory requirement that no province will lose seats by a redistribution. The result has been a slow rise in the size of the House of Commons.

Redistributions normally take place every ten years to take account of shifts in population, and one was due during the rule of the 1993 Liberal government. Faced with the prospect of changes which would increase the number of MPs from 295 to 301, the increase being in Ontario and the western provinces (not a good area for the Liberals), the Liberal government introduced legislation to defer the process until after the next election. This was blocked by the Senate, a popular move which restored some of the Senate’s tarnished reputation.

The redistribution system does have anomalies. Because of the senatorial rule, Prince Edward Island has nearly three times as many seats as its population would justify, and the other Maritime Provinces are advantaged to a lesser degree. Canadian opinion seems to be that these are distortions that can be tolerated in the interests of having all the regions effectively represented in the House of Commons.

The distribution of the seats allocated to each province is done by independent provincial boundary commissions. Each of the commissions is chaired by a judge, selected by the Chief Justice of the province, and there are two members selected by the Speaker of the House of Commons. These two members cannot be MPs. The commissions are intended to produce electoral districts of equal population, but in special circumstances may vary them up to 25 per cent either way. The special circumstances include such matters as density or growth of population, and community or diversity of interests. The drawing of the boundaries and the use of the population discretion naturally excite fierce controversy, particularly from MPs who may criticise the commissions, but cannot overrule them. The process is lengthy, and a redistribution may take three years or more.

The Canadian provinces

In the past, there have been allegations of electoral gerrymandering in some of the provinces and of malapportionment in all of them. There have also been suggestions that provincial parliaments have been slow to make redistributions when population shifts have occurred. By the end of the 1990s, all but one of the provinces had taken some steps to reduce the likelihood of a gerrymander by appointing as chairman of the electoral boundaries commission a judge, a retired judge or (in Quebec) the chief electoral officer, though many feel there is still substantial gerrymandering.

The rural and remote area bias has been the subject of a long-running debate in Canada. The appeal of an equal value for votes, ‘rep by pop’ as it is called, is strong. On the other hand, many feel that the regions need special representation, and in the absence of upper houses the only way this can be done is by a regional bias in the assemblies. The situation was complicated by the adoption in 1982 of a Charter of Rights and Freedoms as part of the Canadian Constitution. As far as voting is concerned, the Charter provides that ‘every citizen of Canada has the right to vote in an election of members of a ... legislative assembly’, that ‘every individual is equal before and under the law and has the right to the equal protection and equal benefit of the law’ and also that the right is ‘subject only to such reasonable limits prescribed by law as can demonstrably be justified in a democratic society.’

A number of disputes over whether electoral boundaries meet the requirements of the Charter have been taken to the courts. In 1987 the Chief Justice of British Columbia held that there was no requirement for absolute voter parity and that the provincial parliaments had the right to permit deviations, but that limits should be set and the permissible reasons for variations should be laid down. It has also been held by the Supreme Court of Canada that the right to vote includes:

  1. the right to cast a ballot;
  2. the right not to have the political force of one’s vote unduly diluted;
  3. the right to effective representation; and
  4. the right to have the parity of the votes of others diluted, but not unduly, in order to gain effective representation, or in the name of practical necessity.

Most of the provinces now follow these rules, and stay within plus or minus 25 per cent of the mean numerical size of electorates. Typical factors to be considered by the boundaries commissions are density of population, accessibility and community of interests. British Columbia and Quebec permit the 25 per cent to be exceeded in ‘special [or exceptional] circumstances’, or ‘when necessary or desirable’, and Saskatchewan permits 50 per cent variance in the north of the province. The effect can be considerable. Moreover, the discretion of the boundaries commissions is limited in some provinces by instructions in the Act that there are to be a specified number of seats in particular areas. In Alberta, for instance, the Electoral Boundaries Commission Act provides for 83 seats in the Alberta legislature, of which 43 are to be allocated to seven cities which account for more than 60 per cent of Alberta’s population. In Newfoundland and Labrador there must be special consideration given to the aboriginal people in Labrador, and to those affected by the inaccessibility of the coast of Labrador and the south-west coast of Newfoundland. Difficulty may arise if the population balance changes, for an amending Act would be required, and this might not be forthcoming.

Ontario uses the federal electorates for the provincial parliament, so that they are regularly reviewed every ten years. The maverick provinces are New Brunswick, Nova Scotia and Prince Edward Island, where the legislative assemblies make up the rules whenever there is a redistribution, which is not very often. In Nova Scotia in the 1999 general election 17 939 votes were cast in one electoral district and 6169 in another. Until the 1998 election Prince Edward Island had sixteen dual member electorates, but this was changed at that time to 27 single member electorates. It is clear that in all the Canadian provinces the effectiveness of the provincial parliaments as electoral colleges is reduced by the rural and remote area over-representation, but the distortion seems to be in accordance with the wishes of the voters, or at least accepted by them, though if the distortions are excessive it seems inevitable that there will be legal challenges under the Charter of Rights and Freedoms.

Redistributions generally occur either every ten years or after every second general election, though Quebec has a redistribution after every election. Again, Nova Scotia, New Brunswick and Prince Edward Island are different, for they have a redistribution only when the government thinks one is necessary, which is not very often.


Under the Australian Constitution the number of members of the House of Representatives from each state or territory must be in proportion to its population, with the proviso that no state shall have fewer than five members. To maintain the two-to-one ratio between the House and the Senate, the formula is that the ‘population quota’ is the combined population of the six states divided by twice the combined number of senators from those states. This population quota is then divided into each state’s and territory’s population to calculate the number of members to which each state and territory is entitled.

Early in each parliament there is an examination of the latest census figures to determine whether the entitlement of any state or territory has changed. If it has, the number of members from that state or territory is amended, and a redistribution is required in that state or territory. Otherwise redistributions are carried out every seven years, or earlier in any state or territory where a third of the electorates in that state or territory are beyond the permissible 10 per cent variation from the average.

The Electoral Commission appoints a Redistribution Committee for each of the states, the membership including the Electoral Commissioner himself and the state Surveyor-General and the state Auditor-General (or their nominees or deputies). The committees must draw electoral boundaries so that the number of voters in each electorate is within 10 per cent of the average for the state or territory, and must endeavour to ensure that three years after the redistribution the number of voters in each electorate will be within 3.5 per cent of the state or territory average. The committees hold public hearings, and they must ‘give due consideration’ to such things as the means of communication and travel, community of interests and existing electoral boundaries. The rules for the committees are laid down in the Electoral Act, which of course could be amended by parliament, but otherwise parliament has no role in the timing or result of redistributions. The Electoral Commission’s decisions are final. They cannot be challenged in any court.

The Australian states

There is little modern evidence of gerrymandering, though in a redistribution in Queensland in the 1980s an Aboriginal reserve (whose inhabitants would vote overwhelmingly Labor) was excised from a marginal National Party seat and placed in a neighbouring safe Labor seat. (The government was National Party.) Such cases are rare. Much more common was malapportionment, which was achieved by using a system of electoral zones, with numerically much smaller electorates in the country. The justification for the electoral zones was that it was very difficult for an MP to service a far-flung electorate, and also the (usually unspoken) feeling that primary producers had a ‘stake in the country’ and therefore deserved greater political power. All the states except Western Australia and Queensland have taken action to make their electorates equal in voter population, and to keep them so with regular redistributions. Tasmania has been exemplary in this respect. Its House of Assembly is now made up of 25 members, with five elected by proportional representation from each of the five federal electorates, which are maintained in equality of numbers by the Federal Electoral Commission.

In Queensland, after the long-serving National Party government was ejected in disgrace in 1989, action was taken to reduce the malapportionment. But the 1992 Act did allow for a decrease in the number of enrolled voters in electorates of more than 100 000 square kilometres, which applied to five enormous electorates in Western Queensland. They have only two-thirds of the average number of voters in the other electorates.

Western Australia has a unique problem. The remote communities are as likely to be mining settlements (and therefore Labor voting) as they are to be rural areas (and therefore Liberal voting-the Nationals are weak in Western Australia). There is therefore no real pressure from any party for voter equality. The distances, too, are immense, and vast areas are barely inhabited. The federal electorate of Kalgoorlie, for instance, in order to reach its quota of voters has to cover an area of more than two million square kilometres. State electorates have fewer voters, but some still cover enormous areas. In the 1994 redistribution, Perth electorates had an average of 22 370 voters, whereas remote rural and mining areas averaged only 11 887. It seems likely that electoral zoning will persist in Western Australia for the foreseeable future.

New Zealand

New Zealand used single member constituencies as the sole basis for the House of Representatives until the 1993 election. In the 1996 election the system was changed to be partly proportional, partly single member electorates. There were 65 members elected from single member electorates (five of them Maori) and 55 members elected by nationwide proportional representation. The system of drawing electorate boundaries remained the same. A seven-member commission (from which MPs are excluded) draws new boundaries after each five yearly census. The commission has to aim for equal size electorates, but is allowed a 5 per cent leeway, to take into account existing electoral boundaries, community of interest, communications and topography. Neither Parliament nor the government has any control over the result.

Voting systems

The seven Australian parliaments use preferential, compulsory voting. The other thirteen parliaments use optional voting, and the voting system is first-past-the-post in all of them, except for the 55 members of the New Zealand Parliament elected by proportional representation.

The advantages of first-past-the-post voting are that it is easily understood and the results can be promptly announced. The voters choose a single candidate, and the one who gains the most votes is the winner. The disadvantages of such a voting system are that two or more candidates may take votes from each other, and another candidate who certainly would not have been the preferred choice of the majority of voters may win. At various times the UK has toyed with the idea of preferential voting, but nothing has come of the proposals. Preferential voting overcomes the problems of first-past-the-post, but it is complex and the result of an election may take some time to become clear. If not distorted by party deals, it may allow similar candidates to offer themselves to the voters without risking both of them being losers-a variant of the American primary system. But party organisers do not like such contests, so they rarely happen. The more complex the voting system, the more it tends to be controlled by party organisers. Voters are usually required to indicate their order of preference for all the candidates, and many of the voters, particularly the reluctant ones, may be marking large parts of their ballot papers in blind ignorance. To overcome this, party workers usually distribute ‘how to vote’ cards at polling places, showing their supporters where to place each sequential number on the ballot paper, with the sequence designed to maximise the vote of their party candidate as the preferences are distributed, and disadvantage the principal opponent. Frequently there are deals done between candidates (or usually between the party machines) to exchange preferences. It all increases the power of the party machines.

Compulsory voting

A bill to introduce compulsory voting passed the Australian Federal Parliament in very casual fashion in 1924. At this time Queensland was the only state to have compulsory voting, though compulsory enrolment was general. A backbench government party senator introduced a private member’s bill to bring in compulsory voting for federal elections. Only five senators spoke in the debate and only three MPs when it reached the House of Representatives. No party leaders spoke on it at all, yet the bill was passed, without divisions in either house. The effect of compulsory voting, which is now used in all federal and state elections in Australia, is to relieve party workers of the tasks of inducing voters to enrol and to go to the polling place. Parties concentrate on the swinging voters, which has the effect of pushing the major parties towards the middle ground.

The argument advanced in 1924 for compulsory voting was that if eligible voters were forced to vote, they would have to consider political issues and become better informed voters as a result. There is no evidence that this has happened. Making the vote compulsory when using the complicated preferential voting system has also introduced a random distortion. Between two and three per cent of the voters, uninterested in the result and nicknamed ‘donkey voters’, number their votes from top to bottom of the ballot paper. A much smaller proportion number their votes from bottom to top, and others scatter their numbers around the ballot paper. The scatterers have no effect on the outcome, but there is a considerable advantage for a candidate being higher up on the ballot paper than his or her principal opponent. The names of candidates used to be listed in alphabetical order, and this gave parties a strong incentive to find candidates with advantageous surnames. To overcome this, candidates now draw lots for positions on the ballot paper, but the result is that a significant advantage, which could be decisive in a marginal seat, is given by random chance to whichever of the two strongest candidates draws a higher position on the ballot paper. This defect could be removed by the use of Robson rotation, which is described when proportional representation is considered.

One of the great benefits of compulsory voting is that it prevents pressure groups from having excessive influence. Such pressure groups can often persuade nearly all of their members to vote, and if the total number of voters were only 50 per cent of those eligible, the voting power of the pressure group would be doubled. With compulsory voting, when typically more than 96 per cent of those eligible do vote, the strength of the pressure group is put in proper perspective.

Compulsory voting is not unique to Australia. Other countries which use compulsory voting are Argentina, Belgium, Brazil, Greece, Italy, Liechtenstein, Luxembourg, Singapore and Venezuela.

Proportional representation

Proportional representation is widely used in democracies around the world, but only two of the twenty parliaments we are considering use it for their lower house elections. There are many varieties of proportional representation, but all follow one of two alternatives-the party list method or the quota-preferential method.

Under the party list method, the voter is offered several party lists of candidates from which to choose one list. Seats are usually allocated to the parties in proportion to the number of votes their lists received, and the winning candidates are usually selected in the order in which their names appear on the party lists. This method, of which the best known is that of d’Hondt, is favoured by party organisations, which control the names on the list, the order of election, and the filling of casual vacancies. The voters have no say in any of these matters. The party list method is much used in continental Europe and in South America. A refinement which advantages the major political parties is to require a certain percentage of the vote-5 per cent is used in Germany-before a party is entitled to a seat.

In 1993 New Zealand adopted the party list method for the 1996 and subsequent elections, when the voter chose, on a second ballot paper, the 55 members not elected from single member constituencies. New Zealand adopted the German requirement for a party to get 5 per cent of the vote on the party list, or win a single member constituency, to qualify for consideration for additional party list members. MPs were selected from the party lists so that the total number of MPs (those elected in single member constituencies plus those from the party list) fitted the pattern of the party list voting.

In the quota-preferential method, voters list the candidates in the order of their preference. A quota is set, according to the number of seats to be filled. Each candidate gaining a quota is elected and any surplus votes are distributed according to the second preferences. If all seats are not filled at the end of this process, the candidate with the lowest number of votes is eliminated, and his or her next preferences distributed, and so on, until all the vacancies are filled. Casual vacancies are filled by a recount of the vote, with the vacating member eliminated. Unlike the party list method, this procedure does not require a minimum quota to keep out Independents and tiny parties, for the size of the quota is determined by the number of seats to be filled. The quota-preferential method, which was developed independently by Carl Andrae in Denmark in 1856 and Thomas Hare in England in the following year, gives the voter wide freedom of choice, and is therefore not at all to the liking of the major party organisations. All sorts of devices have been developed to make its outcome more like the party list method. One was to change the method of filling casual vacancies, so that it is filled by the parties, not the voters, thus eliminating the need for a party to nominate more candidates than it can hope to elect. To control the voters, how-to-vote cards are distributed to voters, guiding them how to fill in their ballot papers in accordance with the wishes of their political party. As a refinement of this, there is pressure for voters merely to have to tick a ‘party box’ on the ballot paper, and their ballot paper will be counted in accordance with their party’s wishes.

Tasmania is the only one of our twenty parliaments which uses the quota-preferential method of election for its lower house. To its credit it has resisted much of the pressure from party organisations to reduce the effective choice of individual voters. Casual vacancies are filled by a recount of voters’ ballots, not by a decision of the political party concerned. There is no marking of ‘party boxes’, and how-to-vote cards have been rendered pointless by a system called Robson Rotation. Under this system ballot papers are printed in batches, each batch having the names of the candidates in a different order. The ballot papers are shuffled and distributed at random, so that each candidate gets equal exposure in key positions. Everyone seems to agree that the result is fair. The important point is that how-to-vote cards are useless, because party workers have no way of knowing which version of the ballot paper a voter will receive. The result is that voters wishing to vote for a particular party have the ability to choose which of that party’s candidates they want to see elected. In a way, they are combining the roles of an American primary election and the election for the actual seat.

Nevertheless Tasmania has not been immune from effective party political pressure. In 1998 the number of members from each electorate was reduced from seven to five (and the quota increased from 12.5 per cent to 16.7 per cent) in a blatant, and successful, attempt to reduce the number of MPs elected from minor parties.

A consequence of the electoral system is that Tasmanian assemblymen are even more involved than other politicians in attempting to achieve benefits for their electorates and even busier attending functions and knocking on doors. The reason is that although Tasmania has about the area of Scotland, it has only 330 000 voters, and is very parochial. Tasmanians tend not to follow party tickets, but rather to vote for those they know and like, or at least have heard of. Some Tasmanian state politicians therefore spend more of their time campaigning against their colleagues than against the opposing party.

The only other use of proportional representation for full lower house elections was in New South Wales between 1920 and 1926. The quota-preferential method was used, with five-member electorates in the city and three-member electorates in the country. Unfortunately the system was not well worked out-there was initially no procedure for the replacement of members who died or resigned, for instance-and there was general relief, anyway from the major parties, when the system was dropped. Three Canadian provinces, Alberta, Manitoba and British Columbia used the Hare system of proportional representation for a few multi-member ridings, but proportional representation has not been used since 1956.

Although proportional representation is probably the best way of choosing a representative legislature, it is not so clear that it is the best way of choosing a government. The normal consequence of proportional representation is that no party has an absolute majority, and the smaller parties and Independents are substantially represented. There are often prolonged post-election negotiations-sometimes up to six months-to try to put together a majority coalition. The resultant government might well be one which would not have received a majority of the votes if put to the voters at the time of the election.

Minority governments

If the result of an election is not a decisive win for a party or a coalition, there are several options. A major party can attempt to form a new coalition with minor parties and Independents which would give them a majority, or if this is not possible, to attempt to reach an agreement of support on votes of confidence and budget bills with sufficient MPs to make the government secure. If this also is not possible, a government may soldier on as a minority government, hoping for the best, possibly waiting for a suitable issue on which to call an election.

A good example occurred in Ontario. After 32 years in office the Progressive Conservative government failed to gain an absolute majority at either the 1975 or 1977 provincial elections, but remained in power until it regained its majority in 1981. The other two parties-the Liberals and the New Democrats-were roughly equal in strength, but one or the other always supported Conservative Premier Davis in confidence motions, so the elections were held when the premier wanted them to be held. Premier Davis was the man in the middle and was able, by negotiating sometimes with one, sometimes with the other opposition party, to pass most of his legislation without unacceptable amendments. Indeed, he was sometimes able to use his minority situation to head off unwelcome demands for legislation from zealot supporters.

In the early days of a hung parliament, it is usually fairly easy for a minority government to reach some sort of accommodation with those holding the balance of power, for a very early election would focus attention on the major parties at the expense of minor parties and Independents, who would tend to be blamed for the instability. As time passes, this fear declines.

Minority governments have been not uncommon in the four national parliaments in the 30 years since 1970. There have been none in Australia during that period, unless one counts the minority Fraser Government of 1975, appointed by the Governor-General after he had dismissed Labor Prime Minister Whitlam. After Fraser had been in office for only a few hours an election was called, which he won by a landslide.

In Canada federally there have been two minority governments, both in the 1970s. Elected in 1972, a minority Liberal government lasted for sixteen months before being defeated on a vote of confidence. The Liberals easily won the resultant election. In 1979 a minority Conservative government lasted for seven months (not facing parliament for the first five) before being defeated on a vote of confidence. The Liberal opposition won the election.

In the 1993-96 Parliament in New Zealand, when it was known that MMP voting would be used for the next election, some members broke away from the major parties and formed splinter groups. The National Party lost its majority and continued until the 1996 election as a minority government, with a guarantee of basic support from a minor party, the United Party. Parliament passed out of government control, and a Business Committee, with the Speaker as chair, was set up to take control of parliamentary business. Some private members’ bills were passed, including one which was opposed by the government at all stages. In the two general elections held in New Zealand since MMP was introduced, no party has held an absolute majority, but on each occasion it has proved possible for the largest party to form a coalition with one or more minor parties to give it a majority, though in August 1998 the National Party’s coalition partner, the New Zealand First Party, walked out of the coalition, leaving the National Party as a minority government until the election in the following year, which was won by the Labour-Alliance Coalition, but it too was a minority government.

At Westminster there have been three minority governments in the past 30 years, under Wilson from February to October 1974, under Callaghan from March 1976 until March 1979, and under Major in the last months of the 1992-97 Parliament. Wilson survived for a few months because none of the non-government parties wanted another election so soon, and Callaghan because of an agreement with the Liberal Party (the ‘Lib-Lab Pact’) which kept Labor in office as a minority government, but the Liberals eventually ended the agreement and voted with the Conservatives and Scottish Nationalists to throw the Labor government out. After the 1992 election Major’s Conservatives had a small majority over the Labour Party, but this was whittled away by defections and by-election losses. He was never in serious danger of defeat on a confidence motion because of the support of the Conservative defectors and the Ulster Unionist Party on such motions, but he had no reliable majority for his legislation, particularly on the deeply divisive European Union issues.

The only examples of minority governments in the Canadian provinces during this period were the 1975-81 and 1985-87 Ontario governments, 1988-90 in Manitoba and 1998-99 in Nova Scotia. The 1975-81 minority governments in Ontario have already been described, but the 1985-87 government was nearly as interesting. After the 1985 election the Progressive Conservatives were the largest single party, but did not have a majority. Two weeks after the start of the new Parliament the other two parties combined to defeat the Progressive Conservative government’s motion for an Address in Reply to the Speech from the Throne. The premier resigned without asking that Parliament be dissolved. The Lieutenant-Governor than accepted a formal ‘Accord’ between the Liberals and the New Democrats, based on an agreed program, and promising that the New Democrats would ensure the survival of a Liberal government for two years. This they did, and in the 1987 election at the end of the two year period the Liberals won an overwhelming victory.

Five of the six Australian states have had minority governments at some time during the past 30 years. The only exception is Western Australia. As might be expected with proportional representation, Tasmania has had the most, with minority governments in 1969-72 (Liberal), 1981-82 (Labor), 1989-92 (Labor) and 1996-98 (Liberal). Minority governments in the other states occurred in New South Wales (1991-95), Victoria (1999-2003), South Australia (1975-77, 1990-93 and since 1997), and Queensland (1996-98).

Some of the ways minority governments won office and survived are interesting. In Tasmania in the 1989 election the Labor Party won only thirteen seats in the 35-member House of Assembly, with the Liberals winning seventeen and the Greens five. The Labor leader managed to do a deal with the Green Party so that he could form a single party minority government. The price was high. The Greens had 100 conditions for the accord with the Labor government, including the cancellation of a major new project to use waste timber from sawlog operations, the setting up of new parliamentary committees, a register of the pecuniary interests of members, public disclosure of election donations, and the ‘abolition of subsidised liquor to members’. The Greens were also given equal status with the Liberal Party in the chamber, and were guaranteed consultation on legislation and public service appointments. The arrangement lasted for a little over two years, but the parties then parted acrimoniously. The Liberals won an absolute majority in the resultant election.

In South Australia, after the 1990 elections the Labor government had only 22 seats in the 45-member lower house. There were three Independent members, all with Labor Party backgrounds, and Premier Bannon arranged that one Independent should be Speaker, and another should be Chairman of Committees. In 1992, when Labor numbers had fallen to 21, the third Independent was offered the chair of the Economic and Finance Committee. When Bannon resigned in 1992, his successor gave ministries to two of the Independents, and left the third as Speaker. ‘Independent’ had taken on a new meaning.

There are also sometimes ‘minority oppositions’, when no party is clearly entitled to the role. There are considerable advantages in being the official opposition, for the Leader usually receives the same benefits as a minister, as well as extra funds to run the Leader’s office. In the House, the opposition asks the first question during question time, and has the opportunity to respond first to the budget and to ministerial statements. As well, the chairs of important committees may fall to the official opposition.

But what is to be done when two non-government parties are of identical size, and have no wish to go into coalition? This happened after the election in Nova Scotia in 1999. The Progressive Conservatives won a majority government with 30 seats, and the non-government parties had eleven Liberals (the previous government) and eleven New Democrats (the previous opposition). The Speaker decided that there should be no official opposition, and that the procedural advantages in the House should be alternated between the Liberals and New Democrats. The financial benefits would be split between the two parties. The Speaker’s decisions seem to have been accepted.

Public financial support to political parties for election campaigns

United Kingdom

In the UK there is no public financial support to the political parties for election expenses, though each side in the 1975 referendum on membership of the EU received 125 000. Limits are put on how much each candidate may spend in an election campaign, though it is a matter of judgement as to what constitutes an election expense, for the law has not been tested since 1929.


In Canada there is no annual funding of registered political parties at the federal level, though it is done in the provinces of New Brunswick, Prince Edward Island and Quebec. On the federal scene, candidates who receive at least 15 per cent of the valid votes cast in their ridings in a federal election or by-election are entitled to a reimbursement from the government of 50 per cent of their election expenses. All candidates are entitled to a full refund of their one thousand dollar deposits. Registered political parties who obtain at least 2 per cent of the total valid votes cast, or 5 per cent of the valid votes in the ridings where they have candidates, have the right to a reimbursement of 22.5 per cent of their election expenses. The Income Tax Act provides for tax credits to individuals or organisations for financial contributions to candidates and registered political parties up to a maximum of five hundred dollars in any one calendar year.

Most of the provinces supplement these arrangements; the only ones who do not are Alberta and Manitoba. Even in the provinces which do provide support it varies quite markedly, in some cases providing support for the general expenses of registered political parties, while in others the support is limited to provincial election expenses. In British Columbia, for instance, registered political parties may claim tax deductions for contributions, but there is no particular assistance for electoral expenses. In Saskatchewan, if a candidate in a provincial election or by-election receives not less than 15 per cent of the valid votes, the candidate receives an amount from the government which is equal to one-half of the eligible election expenses, though there is a limit as to how much is allowed. In Ontario there is partial reimbursement of the election expenses of candidates in elections or by-elections if they receive 15 per cent of the valid votes, and registered political parties may have part of their electoral expenses reimbursed if they receive an aggregate of 15 per cent of the valid votes in the electoral districts in which they have candidates.

During a federal election every broadcaster is required to make 62 hours of airtime available for purchase by registered political parties. The Broadcasting Arbitrator allocates the available time among the registered parties based on their performance at the last general election.


In Australia, the first federal election campaign for which there was public funding for the parties was in 1984, under the Hawke Labor Government. To qualify for the funding a candidate or Senate group must obtain 4 per cent or more of the formal first preference votes in the electorate contested. The scales were initially set at 66 cents for each House of Representatives vote and 33 cents for each Senate vote, but the rates were indexed to inflation, so that following the 1993 election the political parties received nearly fifteen million dollars in public money.

There was a dramatic increase in 1995, when the funding went up to one dollar and fifty cents for each House of Representatives or Senate vote. Using this system (indexed for inflation) the funding payments for the 1998 federal election reached nearly thirty-four million dollars. Two of the states, New South Wales and Queensland, provide similar support. The New South Wales scheme was introduced in 1981, three years before the federal scheme, and was in fact used as the model for that scheme. Queensland introduced a similar scheme in 1994.

There is little support given through the income tax system. Donations of up to one hundred dollars by individuals to registered political parties are tax deductible, and that is it.

New Zealand

In New Zealand the only assistance given by the government to political parties is the provision of some free broadcasting on radio and television during election campaigns. This is given only to political parties which were registered at least three months before the dissolution of Parliament for a general election, and who are contesting at least five seats. The Electoral Commission allocates the available time, on the basis of the votes of the parties at the previous election and the current number of MPs, and also makes financial grants to the parties on the same basis to pay for the production costs of broadcasts and the costs of the broadcast time.

Recall of members of parliament

In 1994 the Legislative Assembly of British Columbia passed an act which enabled the voters to remove a sitting member during his elected term. The Act was passed by the Legislative Assembly, reluctantly, as a result of an overwhelming vote in its favour in a referendum. Forty per cent of voters in the relevant electoral district have to vote in favour of a petition asking for the member’s removal for it to be successful, and a petition cannot be initiated within eighteen months of a member’s election. The initiator of the petition has 60 days in which to collect the required number of signatures, and if the signatures meet the requirement the Chief Electoral Officer declares the seat vacant, and a by-election is held. The recalled member can be a candidate in the by-election.

As might be expected, the recall procedure has been often used for blatantly political purposes. If the government has a narrow majority, the procedure can be used in an attempt to force by-elections which the government might lose. It can also be used to mock the parliament. The first seven petitions fell into these categories, with reasons given for the recall such as ‘we can’t blame everything on El Nino’ and ‘they desire recall so that they may elect someone less boring.’ All failed to reach the necessary number of signatures.

The eighth petition was less trivial. It was established that a member of the Legislative Assembly had written a number of letters to the newspapers, using fictitious signatures, in which he criticised opponents and praised himself. The petition was heavily supported, and the member resigned without waiting for the official dismissal.

None of the other parliaments has followed the example of British Columbia.

The power of parliament to ‘dismiss a ruler’

Of course the role of a lower house as an Electoral College does not end with the choosing of a government after an election, as happens with the American Electoral College. As Bagehot wrote, the British House of Commons ‘lives in a state of perpetual potential choice; at any time it can choose a ruler and dismiss a ruler.’

In fact the potential choice has not been exercised much recently. Since 1970 there has been only one occasion on which the UK House of Commons has removed a government by a vote of no-confidence. The Labour government’s three seat majority (319 out of 635 seats) after the October 1974 election was lost by 1976 in by-elections and defections. The Labour Party was deeply divided over membership of the European Community, and there were serious economic and union problems and difficulties over Scottish and Welsh devolution. The Callaghan Labour Government was kept in office by the thirteen Liberals in 1977 and 1978, but in March 1979 the Liberals joined with the Conservatives and Scottish Nationalists to throw it out. No alternative government would have been possible in that House, so there was no dispute about Prime Minister Callaghan’s request for an election. He lost, to Margaret Thatcher.

In Canada federally there have been two occasions since 1970 on which a government has been ‘dismissed’ by the House of Commons. After the 1972 election, a minority Liberal government survived for sixteen months before being defeated on a vote of confidence. This proved to be a serious error of political judgement, for the Liberals won the consequent election with an absolute majority. In 1979 a minority Progressive Conservative government lasted for seven months (not facing the parliament for five of them), sustaining itself with thirteen billion dollars’ worth of special warrants, which do not require parliamentary approval. The government was voted out soon after meeting the House of Commons, and the Liberals easily won the ensuing election.

There have been no ‘dismissals’ since 1970 by the Australian or New Zealand Houses of Representatives.

Changes of prime minister

While the lower house has the power, even if little exercised, to ‘dismiss a ruler’, it normally has no say in the choosing of a new prime minister should this become necessary before the parliament’s term is up. This important task is left to the majority party.

United Kingdom

Since 1970 there have been two changes of prime minister, other than as a result of a General Election. On the Labour side Callaghan replaced Wilson in April 1976, and John Major replaced Margaret Thatcher as Conservative prime minister in 1990. Wilson retired voluntarily, and his successor was elected by the Parliamentary Labour Party, a system which was used from 1922 to 1981. At a special conference in January 1981 an election system for the Leader and Deputy Leader was adopted by which they should be re-elected each year with 30 per cent of the vote allocated to the parliamentary party, 40 per cent to the trade unions and 30 per cent to the constituency parties. This was changed in 1993 to allot one-third of the votes to each of the three groups.

Although Wilson went quietly, Margaret Thatcher had to be forced out. The Conservatives were doing poorly in the polls, there had been resignations of important ministers, including the deputy prime minister, Sir Geoffrey Howe, and she was challenged for the leadership by Michael Heseltine, a former Defence Secretary. The Conservative leader is chosen by the parliamentary party, by a body called the 1922 Committee. When the Conservatives are in opposition the 1922 Committee includes the entire parliamentary party, while in government only backbenchers are eligible for membership. If there is no clear winner on the first ballot (defined as being 15 per cent clear of the next candidate), the election goes to a second ballot. If no one gets more than 50 per cent of the votes, it goes to a run-off between the top candidates. Thatcher won the first ballot, but not with the 15 per cent margin over Heseltine which she needed to be proclaimed leader. She withdrew from the second ballot, new nominations were called, and it was won by one of her supporters, John Major. He defeated Heseltine and Hurd, but did not get 50 per cent of the votes. Heseltine and Hurd then withdrew from the race, and Major was proclaimed leader, although by Conservative Party rules there should have been a third ballot. In the cases of both Wilson and Thatcher the election of the new prime minister was made by the party members in the House of Commons. From the point of view of responsible government, it should be noted that the House of Commons as a whole was in no way involved in either of these proceedings, not even being asked to give a formal vote of confidence in the new prime minister.

The rules for a challenge to an incumbent Conservative leader were changed after the 1997 election defeat, and 15 per cent of the parliamentary party must now endorse such a challenge before an election can be held.


Because of the method of choosing a party leader it is difficult to replace a Canadian prime minister in office. In both the Liberal and Progressive Conservative parties, the parliamentary leader is chosen, not by the parliamentary party, but by a national convention. The conventions consist for the most part of delegates elected at public meetings in the various ridings, with the addition of senators, provincial assemblymen and some appointed delegates-at-large from the provinces. The parliamentary party has no role in the matter. The only prime ministers to be replaced in office since 1970 were Pierre Trudeau in 1984 and Brian Mulroney in 1993. Both retired voluntarily.


Since 1970 two Australian prime ministers have been replaced, apart from those who lost office as the result of electoral defeat. John Gorton, of the Liberal Party, took office in 1968, and soon alienated some of his colleagues by a high-handed and sometimes lackadaisical attitude to administration, and an open liking for centralisation. The leadership of the Liberal Party is decided by the Liberal members of the two houses, and in 1971 Gorton was in trouble, facing the probability of a no-confidence motion in the House, and with his deputy, William McMahon, industriously trying to organise a coup. A motion of confidence in Gorton was moved in the Liberal Party room. The ballots were secret, but, by party decision, were counted on the table in front of the prime minister, and the vote was tied. When a recount confirmed the tie, Gorton used his casting vote against himself. In many ways this was typical of Gorton-dramatic, noble, but ill-founded, for the chairman of a meeting has a second and casting vote only if the rules of the meeting give him one, which the rules of the Liberal Party’s meeting did not do. The MPs, other than the Liberals, first heard that their new prime minister was William McMahon from the media.

During the Hawke Labor Government from 1983 to 1991, the Treasurer was Paul Keating. Keating was ambitious, but had no ambition to be leader of the opposition. He wanted to be prime minister before the life of the Labor government had expired. In 1988 Hawke promised Keating, his deputy leader, in the presence of two witnesses, that he would resign in favour of Keating at a decent interval after the next election. In return, Keating was in future to come to Cabinet meetings on time and was to be polite to his Cabinet colleagues.

Labor won the 1990 election, but Hawke showed no sign of honouring his commitment to resign after a decent interval. (Whether Keating honoured his two commitments has never been made clear.) Hawke believed that he alone could win the next election, his fifth in succession, and his promise to Keating was void. Keating organised a leadership challenge in the Labor parliamentary caucus (comprising both MPs and senators) in June 1991, but lost by 66 votes to 44. Keating retired to the backbench.

Things began to go badly wrong for Hawke. Keating’s successor as Treasurer, John Kerin, was not a success, and was removed by Hawke in December 1991. Unemployment rose each month, and the economy was dormant. In November 1991 the opposition released a dramatic policy statement, proposing major economic reforms. The response of Hawke and Kerin was grossly inadequate, and Labor members were desperately missing Keating’s force and power of analysis. Hawke’s support was evaporating, and in December he resigned, leaving Keating to take over.

New Zealand

Since 1970 prime ministers in office have been changed five times-twice by resignation, three times by party room coups. Both the resignations were of long-serving prime ministers who had established their deputies as their successors. The three coups were dramatic. As in most of the other parliaments, the leader is elected by the parliamentary party. David Lange (Labour) became prime minister in 1984. For the first three years he worked apparently amicably with his reformist finance minister, Roger Douglas, as New Zealand embarked on a program of radical economic reform, but he eventually became worried at the pace of change, rising unemployment and departure from traditional Labour principles. There was also probably an element of pique at playing second fiddle to Douglas. In any event, the two fell into open conflict, and Lange dismissed his finance minister. Six months later the Labour caucus, which elects the ministry, voted to reinstate Douglas in the Cabinet, against Lange’s wishes. Although it is probable that most of the caucus wanted Lange to continue as prime minister, his position was untenable and he resigned.

It was August 1989 and an election was due in little more than a year. The caucus chose Geoffrey Palmer as the new prime minister, but Labour’s electoral prospects continued to deteriorate as unemployment rose. In desperation, the caucus replaced Palmer with the more aggressive Michael Moore, but it was too late, despite some last minute dramatic changes of policy. The Labour government was trounced.

The winning prime minister was the National Party’s James Bolger, and he lasted for seven years. There was however growing dissatisfaction with his leadership by 1997, and while he was in Britain at a Commonwealth Heads of Government meeting a coup was organised. The fifth-ranked cabinet minister, Jenny Shipley, had the numbers and threatened to force a vote at the regular caucus meeting after Bolger returned. Eventually, to avoid the political uproar of such a contest, he agreed to resign, but after a few weeks to make it more dignified. On 8 December 1997 Jenny Shipley was sworn in as New Zealand’s first woman prime minister.

As in the other countries, the New Zealand Parliament played no role in the change of prime minister.

How long a term?

If the lower house is to perform efficiently as an electoral college, its membership must be changed at reasonable intervals so that it, and the government it chooses, are representative of community opinion. On the other hand the intervals must not be too short, for this would make for ineffective government. Of the 139 parliaments listed in 1999 with the Inter-Parliamentary Union in Geneva, only seven have terms of three years while another three have shorter terms, leaving the remaining 129 with terms of four years or more. Of the four countries being considered, Australia and New Zealand have three year terms, Canada, the Canadian provinces and the United Kingdom five years. All the Australian states except Queensland now have four year terms. The other countries to have three year terms are Congo, El Salvador, Libya, Mexico and the Philippines.

Three year terms, even if they are completed (and they rarely are in Australia), are really too short. The benefits of many desirable new government programs frequently take some time to become evident, and a government facing an imminent election will almost certainly be forced into short term measures. This rarely makes for good government. The problem is exacerbated because parliaments rarely see out their full term, occasionally because of defeat of the government in the lower house, usually because of the use by prime ministers or premiers of their power to call elections at dates which suit their political advantage.

United Kingdom

Of the fourteen elections in the UK since 1945, only one was held at the end of the term of Parliament. The Parliament elected on 8 October 1959 lasted until 15 October 1964, the first peacetime Parliament to run its full term since 1722. The 1992-97 Parliament lasted until it was within twenty days of its full term, almost certainly because Prime Minister Major was desperately hanging on, hoping that something favourable would turn up. It didn’t. Three of the early elections might be thought to have been justified because there was either a minority government or an unworkably small majority. The Wilson Labour Government was in a minority from February to October 1974. The governments which held very small majorities were Wilson (October 1964 to March 1966) and Attlee (February 1950 to October 1951), although Attlee maintained control of the Commons throughout.

That leaves ten other early elections to be accounted for. The 1979 election was forced by the Commons themselves, but the other nine parliaments were shortened by an average of eleven months by the decisions of the incumbent prime ministers. It seems clear that the motive in each case was to take advantage of what was thought to be a favourable electoral climate which might not last if the parliament went its full term. Of course the prime ministers did not say this. The reasons they gave varied from Anthony Eden seeking a mandate as a newly installed prime minister (1955) to Edward Heath asking the people to choose whether they wanted the trade unions to govern (1974). The people narrowly voted for the unions, or anyway against Heath. One could make a case that every prime minister who was not chosen as a result of a general election should seek a mandate as Eden did, but no other prime minister similarly chosen has done so, and Eden did not really want a mandate. He wanted an extended period of secure government.


Since 1945 the average life of a Canadian Federal Parliament has been little more than three years. Even when the government had an absolute majority, no parliament saw out its full five years. With the exception of the 1988 election, there is no evidence of any motive other than party political advantage in the calling of the early elections. The 1988 election was a quasi-referendum on the free trade treaty with the United States, after the Senate threatened to hold up the bill until an election was held. It may have been a mere coincidence that the election was called when the government was leading in the polls for the first time in nearly three years.


The maximum term of the House of Representatives is three years, the term starting from the first sitting of the House after an election. Only four of the 21 parliaments since the Second World War have run to their full term or near it. These were the parliaments of 1946-49, 1969-72, 1990-93 and 1993-96. It is no coincidence that the government was in three cases defeated in the election, and in the fourth (that of 1993) won a surprise victory. Public opinion was running strongly against the government of the day, which held on grimly to the end, hoping that something would turn up.

The motive of a prime minister in calling an early election is always party political advantage, whatever he may say publicly. An Australian prime minister has, however, a unique problem in exploiting this. Senators have fixed six year terms, with half retiring every three years, and the Senate election must be held in the last year of the retiring senators’ term. It is usual for the elections for both houses to be held simultaneously, for a separate Senate election is treated by the voters as a by-election, and the government vote suffers. A prime minister will think very carefully before he calls too early an election.

He has a way out. The Australian Constitution provides for a method of resolving legislative deadlocks between the two houses. If the Senate rejects or unacceptably amends a bill passed by the Representatives, and if after an interval of not less than three months the same thing happens again, there may be a dissolution of both houses. If the deadlock persists after such an election it may be resolved by a joint sitting of the two houses. Before 1951 the double dissolution procedure was used only once, but it has been used much more since then. In 1951 there was a double dissolution over the Senate’s failure to pass a bill to dissolve the Communist Party. This was probably the only genuine use of a double dissolution as a means of resolving a legislative deadlock, though even then there was political advantage seen in fighting a premature election on such an issue. The 1974 and 1975 double dissolutions came about because of the failure of the Senate to vote supply, though the ostensible grounds were deadlocks over other legislation.

It seems that prime ministers now view double dissolutions not as a means of resolving deadlocks over legislation but rather as a method of removing a restriction on the calling of elections whenever they choose. There is no obligation to call an election within any prescribed period after a deadlock has been established, and a government can, if it wishes, store up one or more deadlocks for exploitation at a convenient moment. There is, however, one problem a prime minister must consider. At a double dissolution, all of the twelve senators from each state are up for election and, with quota-preferential proportional representation, the quota for election is only 7.7 per cent. This offers a real chance for minor parties-the Australian Democrats, environmentalists, anti-nuclear groups and so on-to elect senators who may hold the balance of power in the Senate. A prime minister must consider who such groups would favour. Generally they are more likely to favour Labor than the Liberal-National coalition.

Turning back to the nineteen parliaments since 1945 which have not run their full term or near it, if one disregards the five double dissolutions called to resolve legislative deadlocks between the two houses, and the 1963 election which could be argued, with some difficulty, to have been necessary because of the government’s narrow majority, one is still left with thirteen short term parliaments. Three lasted less than two years, one as little as fifteen months. The average was 29 months, seven months shorter than the prescribed period. It is difficult to see any justification for this constant pattern. It seems to be generally agreed that three year parliaments are really too short, yet Australia is doing much worse.

The complex voting system used in Australia poses another problem. The new Parliament cannot meet until the writs are returned, and there is inevitably a substantial gap between polling day and the return of the writs to permit the counting of the votes (including absentee and postal votes) and the distribution of preferences and quotas. Since 1970 the average interval between polling day and the first meeting of the new federal Parliament has been 58 days. Limits on the intervals between the issue of the writs, closing of nominations and polling day, and between polling day and the return of the writs, are set by an act of Parliament, but in practice the nation may be without a parliament for four months, and the responsible government with no one to be responsible to. This system subverts the intention of the Constitution to keep to a minimum the time the nation would be without a parliament, but the authors of the Constitution did not envisage the voting systems now employed.

This problem should be kept in mind by politicians considering the theoretical advantages of various voting systems. An extreme example occurred during the first election of the Legislative Assembly of the Australian Capital Territory in 1989. A modification of the d’Hondt system of proportional representation was used. The ballot paper was a metre wide, and it took more than two months to count the votes and distribute preferences. The result was indecisive.

New Zealand

New Zealand has had the most orderly electoral college of any of the countries, states and provinces we have considered. With two exceptions the elections have been called meticulously at regular intervals in November of each third year. The two exceptions were the 1951 and 1984 elections. The 1951 election was called only 21 months after the previous election because of a disastrous waterfront strike which had crippled the economy, and the election in 1984 was called four months early because of a threatened defection by a National Party MP. Until 1986, however, the New Zealand Parliament did not have to meet at any prescribed time after an election. The only limit was that it had to meet before 30 June, when supply runs out. It was quite common for Parliament not to meet for six months after the election, so that the actual life of the Parliament was often less than two-and-a-half years. The Constitution Act of 1986 now requires the Parliament to meet within six weeks of the day appointed for the return of the election writs.

Fixed terms for lower houses

Except in New Zealand and Western Australia, prime ministers and premiers have been ruthless in their exploitation of their power to call elections whenever they see a window of political opportunity. The power to determine the date of an election is obviously a great advantage to an incumbent, but why should such an advantage be given? It has a downside, too, for a year or more before the end of a parliamentary term speculation starts about when the election will be held, the uncertainty frequently causing economic damage. Often the speculation has been stirred by the prime minister or premier, who then calls an early election to end the uncertainty.

So far the only parliament to adopt a fixed term is New South Wales. In 1992 the New South Wales Parliament passed an act adopting a fixed four year term for its Parliament, and this was endorsed by three-quarters of the electorate at a referendum in 1995. The Parliament can be dissolved earlier only if a motion of no confidence is passed, or the Legislative Assembly rejects or fails to pass a supply or appropriation bill, and a government which has the confidence of the Legislative Assembly cannot be formed within eight days. A similar act was passed by Tasmania for the 1992-96 Parliament only, and it has not been re-enacted. Because the upper house-the Legislative Council-has the power to block money bills, the Tasmanians also provided for a dissolution of the lower house if either house failed to pass the necessary funds for the ‘ordinary annual services’ of the government, thus continuing to make it possible for the upper house to force a premature election on the lower house.

Why should not all our parliaments have fixed terms? Such a step would remove a great deal of unnecessary uncertainty, reduce the excessive frequency of elections, and take away an unwarranted advantage given to incumbent governments. There would of course have to be an early election if no government possessing the confidence of the lower house could be formed.

Objections have been raised to the fixed term concept, but they seem to have little substance. Bagehot thought that a prime minister’s power to order an election was essential to party discipline, and the idea was put in modern terms by David Butler when he wrote that ‘if government MPs know that a defeat on a major issue will lead the prime minister to dissolve parliament, they have a powerful incentive to ensure to vote loyally on key occasions.’ There is no evidence that any modern prime minister has called an election for such a reason, and any such threat would be simply not credible. It is true that Harold Wilson, in March 1967, warned the Parliamentary Labour Party that votes against the government might result in their ‘dog licences’ not being renewed, and the chief whip later told a meeting of the parliamentary party that failure to carry the Industrial Relations Bill would mean a dissolution. Wilson claims that he gave the chief whip strong advice to the contrary, which is reasonable, because whatever bluff a prime minister might use to try to discipline party members, including calling for a vote of confidence, a premature election with the party in a state of turmoil would not be an attractive option. Instead, finding some Labour MPs bitterly opposed to the legislation, Wilson modified it substantially, as many prime ministers have done in the past and many more will do in the future. A desperate prime minister might call for a vote of confidence in the House, which if lost would result in an election, but this possibility is covered by the fixed term concept.

The second argument advanced against fixed terms is that the election date is not flexible, and may fall on an inconvenient date, perhaps during a grave crisis. This is undoubtedly true, but the danger can be exaggerated. In Australia, for instance, if elections had been held at three year intervals after the first election on 31 March 1901, on no occasion would the election date have been politically inconvenient, whereas at least one date chosen by a prime minister was very awkward. Australia was in the middle of an election campaign when the First World War broke out.

It is true that as a parliament’s term draws to a close, great events may be happening which would render an election most inopportune, but this would apply whether there was a fixed or flexible term. The New South Wales fixed term can be cut short by up to two months if it would fall ‘at the same time as a Commonwealth election, during a holiday period or at some other inconvenient time’. National governments may have more serious problems. In both World Wars Britain avoided elections by extending the life of the House of Commons, which can be done by a simple act of Parliament. Canada has a constitutional provision that ‘in time of real or apprehended war, invasion or insurrection’ the House of Commons may, by a two-thirds majority, extend the life of the Parliament indefinitely. Under New Zealand’s Electoral Act an extension of the life of the Parliament can be obtained either by referendum or a vote of three-quarters of the members of the House of Representatives. There is no limitation on the duration or purpose of the extension. To extend the life of the federal Parliament in Australia would require an amendment to the Constitution. This would have to be passed by a referendum which, in the crisis situation envisaged, would probably be nearly as disruptive as an election campaign.

Although it is clearly desirable that national parliaments should have some power to defer elections in time of crisis, there should be some control over the method of extension and the acceptable purposes. The Canadian model seems suitable, though the New Zealand three-quarters majority is better than the Canadian two-thirds, for it is far from unknown for a government to have two-thirds of the members of a lower house. It is important that both government and opposition should be in favour of the extension. It should be noted that these controlled arrangements for extending the life of parliament are desirable whether the normal term of the parliament is fixed or flexible.

The only other apparently cogent argument against fixed terms is that they would prevent a government from obtaining an electoral mandate for a substantial change in the policy on which it was elected. As an opponent of fixed terms put it, a government ‘may wish to make a serious change of policy, but may wish to seek the endorsement of the electorate, surely not an undemocratic thing to do.’[25] This sounds admirable, but is not in fact what happens. If a government makes an abrupt change of policy and the prime minister calls an election, it is because he believes the issue will help him to win the election. If there were an abrupt change of policy which was not well received by the public, no prime minister would contemplate holding an immediate election. He would hold on, hoping that the benefits of the new policy would be apparent by the time an election was inescapable.

To look at the UK, dramatic developments or changes of policy which might be claimed to have required a mandate include the dismantling of the Indian Empire (1947), the decision to produce an atom bomb (1948), the involvement in the Korean War (1950), Suez (1956), the Falklands (1982), and entry into the European Economic Community (1971). A mandate was not sought for any of these at the time of the decision. Moreover, an election is not a very satisfactory method of determining public opinion on a single issue. Other issues have a tiresome habit of intruding, and sometimes the issue on which an election is ostensibly called is barely mentioned during the campaign. An example is the 1987 federal election in Australia which was called to resolve a deadlock over the proposed introduction of a national identity card. The identity card played no significant role in the election campaign. The Labor government was returned, but this was in no sense an endorsement of the identity card, which remained very unpopular with the public.

If one really wants to know what the public thinks on a particular issue, public opinion polls are available. If one wants a more formal expression, referendums are available. Both public opinion polls and referendums have limitations, the principal one being the difficulty of explaining complex issues to some of the voters, but as measures of public opinion on single issues they are far superior to general elections.

One other possible ground for arguing against fixed terms must be mentioned, if only to be dismissed. There have been many occasions when a prime minister or premier has been replaced, because of death, resignation or a party coup. Should the voters not have the right to register their approval or otherwise of the replacement at a general election? The only replacement British prime minister to call an election was Anthony Eden, in 1955.[26] He claimed to be seeking a mandate, but in fact was seeking to exploit the electoral honeymoon which is usually granted a new prime minister. The same thing has happened twice in Canada, in elections called in 1968 by Pierre Trudeau, and in 1984 by Trudeau’s successor, John Turner. Trudeau judged the electoral climate correctly, but Turner was decisively beaten.

Under the present system, the decision whether or not to call an election has nothing to do with principle, but is a matter of political judgement. If a new prime minister fears his electoral honeymoon is not sweet enough to ensure an electoral victory, there is no way an election would be called. There seems to be no general feeling that a new prime minister should have to ask the voters for a personal mandate, and that being so, it is surely wrong to permit an incoming prime minister to cut short the term of parliament to exploit a political opportunity.

The final argument against fixed terms seems to consist of setting up a straw man and then knocking him down. The straw man is the claim that a fixed term could not work because of the possibility of the lower house being unable to agree on a government. In fact, those who advocate fixed terms agree that there must be an election if a government with the confidence of the lower house cannot be formed. Such dissolutions would be limited, for a defeated prime minister or premier would have no right to a dissolution if he lost the confidence of the lower house, for the head of state would have the duty to see if another government which possessed the confidence of the house could be formed. This is the established position in Australia, both federally and in the states. It seems to be the position in the United Kingdom, New Zealand, Canada and the provinces, though it has not been tested in any of these parliaments since 1926. It would be helpful if it were clearly spelt out, in each Constitution or Electoral Act, that a dissolution after a government lost the confidence of the lower house would be granted only if no other government could be formed.

Upper houses forcing elections

There were seven cases in the twentieth century of upper houses forcing premature elections, all but one of them in Australia, the only place where upper houses still have such power.

United Kingdom

The behaviour of the House of Lords with regard to the 1909 budget, and the consequences of its actions, have already been described in Chapter 2.


Although the Canadian Senate technically has the power to reject a budget, such action has never been contemplated, anyway in modern times. The lack of prestige of the Senate and its unelected character mean that it would be almost suicidal for it even to discuss such an idea seriously. Besides, the Senate could not force an election. The Canadians have destroyed a vital part of responsible government by passing the Financial Administration Act, by which the Governor-General may, on the advice of the government, issue special warrants authorising expenditure not approved by Parliament. So, if the Senate blocked supply, the government could simply prorogue Parliament and finance itself by special warrants.

New Zealand

The New Zealand Legislative Council was abolished in 1951, but it was moribund for some time before that and there was never any suggestion of it blocking supply in order to force an election.


There were two occasions during the 1970s on which the Australian Senate interfered with the electoral-college role of the House of Representatives, by refusing to pass supply unless an immediate election was held. The financial year runs from 1 July to 30 June of the following year. At that time the budget for that financial year was not introduced until August (the procedure was to be changed in 1994), and although the passage of the budget through the Representatives was a formality it was not normally passed by the Senate until October or November, after hearings by estimates committees. As the Australian government, unlike the Canadian, cannot spend any money not approved by Parliament, the government was given an advance before the Parliament rose for the winter recess. The amount was of a size sufficient to see the government through for about five months, by which time the budget should have been passed and the amount advanced absorbed in it.

In April 1974 the Whitlam Labor Government had been in office for seventeen months. It faced a hostile Senate, where the balance of power was held by the Democratic Labor Party, a right wing breakaway group from the Labor Party and a ruthless opponent. The Senate had twice rejected six government bills, giving ample grounds for a double dissolution. When the pre-budget supply bills were presented they were threatened with rejection unless Whitlam agreed to an immediate election. Whitlam accepted the challenge, obtained a double dissolution and won the election, but only narrowly. He did not win control of the Senate.

Eighteen months later, with the Whitlam Government in even worse trouble, the Senate deferred consideration of the 1975-76 budget unless Whitlam agreed to a general election. Whitlam decided to tough it out. Supply would last until the end of November, after which government administration would be in chaos. He did investigate the possibility of borrowing from the banks to carry on the business of government, but such borrowings would certainly have been unconstitutional, and nothing came of the plan.

The Governor-General, Sir John Kerr, was in an awful dilemma. He had been advised by the Chief Justice (Sir Garfield Barwick) that he had the constitutional power to dismiss the prime minister, but such power had never been exercised by any of his predecessors. Whitlam made it clear to Kerr that he would never resign or advise an election for the House of Representatives or a double dissolution (the Senate had already given the grounds for a double dissolution), and that the only way an election could be obtained would be by his dismissal, so Kerr did just that. He dismissed Whitlam and installed Fraser, the leader of the opposition, as caretaker prime minister, on the understanding that he would obtain supply and recommend a double dissolution, though whether the Governor-General has the power to impose pre-appointment conditions on an incoming prime minister that he will tender certain advice is very doubtful.

Kerr has been much criticised for his dismissal of Whitlam, though in his defence it must be asked whether a Governor-General, faced with an apparently insoluble political confrontation which was going to result in administrative chaos, did very wrong in asking the voters what they wanted. Very clearly the voters wanted to be rid of the Whitlam Government. On the other hand, the determination of the opposition senators was weakening as public opinion moved sharply against them, and it is most unlikely that they would have held firm as social chaos developed. Whitlam would have. Besides, Kerr’s plan should not have worked. It depended on Fraser being able to secure the passage of the budget through the Senate as an essential preliminary to a dissolution. Whitlam failed to tell his Senate ministers of his dismissal. On the other hand, the Liberal and National Party senators were fully aware of the situation. When the Senate met at 2 pm on 11 November 1975 the Labor Senate Leader almost immediately moved the adoption of the budget as a matter of urgency, and to his astonishment in four minutes it was passed. Of course by now it was Fraser’s budget, and Fraser had his key requirement. The Coalition easily won the election but there was unprecedented bitterness.

The other four supply-blocking incidents occurred in the states, but there have been none in the past 50 years. It is difficult to think of anything to recommend such actions, or any motive other than the desire for political power. The problem is how to dissuade upper houses from forcing premature elections without destroying themselves as effective legislatures. The New South Wales Constitution has a unique provision by which the upper house cannot reject or amend any bill dealing with ‘the ordinary annual services of the government’. There are some problems with the New South Wales approach, which will be dealt with when upper houses are discussed in Chapter 8, but there is no doubt that it is totally effective in preventing the upper house from forcing a premature election.

The Victorian and South Australian parliaments have taken a different approach by adopting a partially fixed term. When their constitutions were amended to provide for four year parliamentary terms, the power of the premier to ask for an election during the first three years of a parliament were effectively limited to circumstances where the lower house had passed a vote of no confidence, or where there was a legislative deadlock between the two houses. Nothing was done directly about the blocking of supply, but it is now a much less attractive option during the first three years of a parliament. It must be assumed that any future government would adopt the Whitlam ‘toughing it out’ tactics, but it could not be assumed that a governor would act like Sir John Kerr. As the state drifted towards administrative chaos, the opposition would have to bear the political odium, which would be compounded if the only way there could be an election was for the government to pass a vote of no confidence in itself. In the fourth year of a parliament, with an election pending, an opposition scenting victory would be extremely rash to risk throwing it away by blocking supply in the Legislative Council. It seems most unlikely that any future Victorian or South Australian oppositions will block supply, no matter how unpopular the government.

There has been less progress on the federal scene. In the bitter aftermath of the 1975 dismissal of the federal government, rational debate has been difficult. Two proposals to amend the Constitution have been put forward. From the left there was a push to remove the Senate’s power over money bills. This would seriously diminish the Senate as a legislature, which is of course the purpose. From the right came the proposal that if the Senate blocked supply and forced an election for the House of Representatives, the Senate too should be dissolved. Neither proposal has the slightest chance of being adopted. Though the danger remains, for the foreseeable future the Senate is most unlikely to risk exercising its disruptive power again. The scars of 1975 are too deep. Of course, if there were a fixed term for the House of Representatives, it is almost inconceivable that the Senate would ever block supply.

The role of the head of state

There have been some extreme views expressed that the head of state has no discretion, but must do what the prime minister requires. Australian Prime Minister Gough Whitlam claimed, while in office, that the Governor-General unquestionably must act on the advice of his prime minister, with no tolerance whatever.[27] Imagine a prime minister who was clearly beaten in an election refusing to resign until he had faced the House, as he is certainly entitled to do. On being defeated in an immediate vote of no confidence, he asks the head of state for a dissolution and another election which, according to the extremist view, the head of state would have to grant. It is only necessary to state the proposition to reveal its absurdity. The head of state must have some discretion.

The head of state must also have considerable discretion in deciding when a government has lost the confidence of the lower house. Whom should the head of state then invite to try to form an alternative government? To whom should the head of state grant the dissolution, if one becomes inevitable after the failure of attempts to form alternative governments, for holding office during an election campaign may be a considerable advantage?

These matters will be discussed later, but it should be noted how substantial is the necessary discretion given to the head of state, whether the term of parliament is fixed or flexible.


How well have the lower houses of the various parliaments performed the electoral-college role? Eighteen of the twenty electoral colleges here considered use single member constituencies as the method of choosing all their members. Although the single member constituency system exaggerates swings and usually produces a decisive result, it does not always result in the government desired by a majority of voters. The preferential voting system used in Australia shows this most clearly. Of the 22 governments chosen by the federal House of Representatives between 1946 and 1998, five would not have been the preferred choice of a majority of voters. Although the figures are not so easy to interpret in the countries without preferential voting, it seems that in at least 10 per cent of the elections which resulted in an absolute majority of seats for one party, that party would not have been the preferred choice of a majority of voters.

Nor have the election results always been decisive. There were minority governments in the UK and Canada in the 1970s, and there were occasional minority governments in the Canadian provinces. In the five Australian states using single member constituencies, the pattern of clear majorities which applied in the 1970s and 1980s suffered an abrupt change towards the end of the latter decade, and in the 1990s there were minority governments in four of these five states, with the balance of power being held by Independents. There were no attempts to stitch together post-electoral coalitions, though some policy concessions were made to particular Independents to gain their support. The election of an Independent member as Speaker was also a popular option.

What can the Australian House of Representatives learn from the other nineteen parliaments we are considering in order to improve its performance as an electoral college? The first issue is the electoral system. Single-member constituencies seem to be the best option, though the system is far from perfect and gives a much less decisive result if a multi party contest develops. It is difficult to see a better option in the other nineteen parliaments.

The Australian federal system of drawing electorate boundaries is exemplary, and nothing useful can be learned from the other nineteen parliaments. The Australian system is fair, prompt and free from political delays or interference. On the other hand, it must be acknowledged that too frequent electoral redistributions are likely to undermine the stability of representation. MPs may expend much effort in establishing close ties with their voters, only to find their electorate boundaries substantially changed, or the electorate even abolished. This tends to make the party more important than the MP, thus vastly increasing the power of the party machines.

The seven Australian parliaments are the only ones to use preferential voting. The same is true of compulsory voting. There is no evidence that preferential voting gives a more decisive electoral result, or one that better reflects the overall wishes of voters, but it is probably desirable in that it produces MPs who are preferred-or perhaps least disliked-by a majority of their voters.

The life of the Australian House of Representatives, a maximum of three years, is far too short. Only New Zealand and Queensland have a similar term. The remaining five Australian state lower houses have four year terms, and the other twelve lower houses have five years. The Australian term should be increased to at least four years as soon as possible.

The term of the Australian Parliament should also be made fixed, as it is in New South Wales. There is no justification for leaving the Australian prime minister with the power to cut short the term of the House of Representatives to suit his political advantage. The term should be cut short only if no government possessing the confidence of the House of Representatives can be formed. The term should be capable of being lengthened-as can be done in Canada and New Zealand-by a two-thirds or three-quarters majority of the House of Representatives if a national emergency made an election highly undesirable.

There is nothing to be said in favour of the Senate usurping the electoral college role of the House of Representatives. If the House of Representatives had a fixed term, the Senate’s power to block supply in order to force a premature election on the House of Representatives would become unusable, for what would be the point of trying to force an election if there could not be one? Nor is a double dissolution a sensible way of resolving a deadlock over legislation between the two houses. These problems are discussed when the roles of upper houses are considered in Chapter 8.

Previous Page | Contents | Next Page

Back to top