Two Electoral Systems for Westminster

Territorial PR is a new form of proportional representation in which each MP has their own territory, from which they get their votes and for which they are responsible. It is very similar to the present UK system of Plurality in single-member constituencies, but nevertheless it is completely proportional, or rather, it can be adjusted by means of a representation threshold to give any required degree of proportionality. It appears that, of all forms of PR, this system will give the best "MP-constituency link", maximising the opportunity of each MP to maintain contact with the electors who have supported them, and indeed it can be argued to be superior in this respect even to the present Plurality system.

Territorial PR can best be regarded as a modification of the present UK system, one which is designed to make the present system proportional, but with the minimum of change. Indeed, the only change it makes from the present system is this: that any party does not place its candidate in one constituency as it does now, but instead is allowed to place its candidate to stand in a "territory" consisting of a cluster of several adjacent constituencies, as many as the party chooses.

Each party is then allocated seats in proportion to the total of votes which its candidates obtain. But a party may elect only one MP for each territory it has created. Thus each party will be careful to create at least as many territories as the number of seats it is likely to get. If, as is quite likely, a party gets fewer seats than the number of territories it has created, then the two adjacent territories in which it got the fewest votes will be merged into one territory, and one candidate will be elected from it. In this way, the party's number of territories will be reduced down to its number of seats, so that one candidate can be elected from each territory.

Thus the territories of any one party (unlike the constituencies of one party under Plurality) will normally cover the whole of the country. If the party is a large one, it will have a large number of territories, each a small one, consisting of perhaps two or three of the present constituencies. If it is a small party, it will have a small number of territories, and each territory will be large, containing many present constituencies.

A fuller explanation of the rules of Territorial PR, illustrated by a hypothetical example, is given in the appendix three pages below.

What then will be the political effect, of thus changing the present system to Territorial PR? Proportionality of seats to votes will be achieved, but it will be a unique type of PR in which each MP of a party will be responsible for his or her own separate territory. Whatever the size of the party, this territory will contain about a quota of voters who have voted for the MP, or (if the party has put several candidates in the territory) for these candidates including the MP. (A quota is defined as: the total of votes of all parties, divided by the number of seats.) In other words, each MP of the party will have about an equal share of the "constituency" responsibility. Thus whether the party is a large one with small territories, or a small one with large territories, each MP's territory can be regarded as being of about optimum size for the purpose of maintaining contact between the MPs and their supporters. Further, whatever party a person has voted for, they will be represented in the parliament by a single MP of that party who is standing for their local area.

How then does this compare with the situation under the present Plurality system? Under the latter, though each MP is responsible for a constituency-sized "territory" with about a quota of voters in it, many or even most of these voters will not have voted for that MP, and moreover will not be in the territory of any other MP who belongs to the party they support. Thus quite apart from its greater proportionality, Territorial PR has the advantage over Plurality that it enables MPs to maintain better contact with their supporters, that is, it provides better "MP-constituency linkage".

For different reasons, Territorial PR appears superior to the usual forms of PR using large multi-member constituencies. The latter systems place most MPs, that is, those of the larger parties which win several seats per constituency, in constituencies which are too big, much larger than the territories these MPs would have had under Territorial PR. Hence here again, Territorial PR enables these MPs to maintain better contact with their supporters, than they could do under a conventional system, that of ordinary PR.

But how does Territorial PR compare with the Additional-Member System? Under the latter system, one-third or perhaps one-half of the MPs are elected from the party lists and effectively have no territories at all. Also, the constituencies of any one party cover only part of the country, from about half of the country, in the case of a large party, down to none of it, in the case of the smaller parties. In contrast, under Territorial PR, all MPs will have territories. Further, the territories of the MPs of any one party, even a small one, will cover the whole country. Thus whatever party you have voted for, you will be represented in the parliament by a single MP of that party who is standing for your local area. This seems an advantage over AMS, where perhaps most people will be represented in their constituencies by an MP for whom they have not voted, and to whom they might be strongly opposed.

Under Territorial PR, a party may, if it so chooses, put more than one candidate in any one territory. Unlike under Plurality, it will not be deterred from doing so by fear of "splitting the vote". Under Plurality, though a party is allowed to put up more than one candidate in a constituency, it will never do so, since this can only reduce the votes which each one of them gets, and its chance of getting any one of them elected. But under Territorial PR, this splitting of the vote between several candidates will not reduce the number of seats a party gets, unless it actually reduces the party's total number of votes over the whole country. In fact, it seems more likely that, by offering a greater variety of candidates, the party will attract more votes, and so will increase its seats. Thus the party may well choose to put up more than one candidate, in some if not all of its territories.

Thus by way of conclusion, the following points can be made in favour of Territorial PR.

(1) The system is simple for the voter, since from the voter's point of view, in terms of the ballot paper and the method of voting, it is exactly the same as Plurality.

(2) Since in effect it uses one nation-wide constituency, the system can be adjusted, by means of a representation threshold, to give any required degree of proportionality.

(3) Under Territorial PR, each MP has their own territory, from which they get their votes and for which they are responsible. Thus it appears that this system gives the best "MP-constituency link", enabling the MPs to maintain better contact with their supporters, than they could do under other systems of PR, or even than under Plurality.

(4) Unlike the constituencies of Plurality or AMS, the territories of the MPs of any one party, even a small one, will cover the whole country. Hence whichever party a person votes for, they will be represented in the parliament by a single MP of that party who is standing for their local area.

(5) In terms of the voter's choice of candidate, Territorial PR is superior to Plurality, and perhaps comparable with open-list PR.

Territorial PR is thus put forward as a simple way in which the present Plurality system can be converted to proportional representation, with the minimum of change.

A more detailed account of the rules of Territorial PR is as follows. The present single-member-constituency areas can continue to be used by the new system. However, since they will no longer be used to elect a single MP in each one, it seems advisable, in order to avoid confusion, to refer to them not as constituencies, but as "tracts". (The proposal that Territorial PR should use the present constituencies, is made for reasons of convenience only. As will be clear later, there is no need to have a number of tracts which is exactly the same as the number of seats--a greater number of smaller tracts could be used, and would if anything work better.)

A party creates a number of "territories", each of which consists of a cluster of adjacent tracts, and it puts one candidate, or if it so chooses, more than one candidate, into each territory. Any one territory of a party may not overlap another territory of the same party, but otherwise the territories of a party may be as many in number, and as large or as small, as the party decides (and they will, of course, overlap the territories of each other party). However, it can be expected that the party will create at least as many territories as the number of seats it thinks it is going to get, and that it will create them of such size that together they cover the whole country. The decision about the size of territories, and about which candidates and how many candidates are to stand in them, should be taken by the MPs of the party. The procedure suggested is that they elect a committee to draft a territory plan for the whole country, which is finally passed by vote of the MPs of the party, in secret ballot.

In order to be able to fill vacancies which occur during the legislative term, a party is also required to run a list of additional candidates, who are placed on the list in the order in which they are to be elected. Candidates who are standing in a territory may also be placed on the additional list, so that they will be available to fill vacancies, if they are not elected at the general election.

There is a separate ballot paper for each tract, which carries the name of each candidate standing in the tract, together with the name of his or her party. The candidates are placed on the ballot paper in order of the votes of their party in the previous election, the candidates of the largest party being placed at the top. In the case of new parties, the more electors their territories contain, the lower down their candidates are placed. ( Thus even if there are many parties on the ballot paper, most voters will be able to find the candidate they want to vote for without difficulty.)

The elector votes for one candidate, with a simple "X". (Thus the ballot paper, and the method of voting, is exactly the same as in the present first-past-the-post system.) A party receives seats in proportion to the total of the votes which its candidates have obtained, over the whole country.

A representation threshold can be incorporated into the system if it is required, giving no seats to a party which gets less than some fixed percentage of the national vote, such as 5 per cent, or 3 per cent, or whatever.

However many seats a party is entitled to on the basis of its votes, it cannot elect more than one MP for each of its territories. Thus we can expect that each party will be careful to create at least as many territories as the number of seats it is likely to get. If, as is quite likely, a party gets fewer seats than the number of territories it has created, then those two adjacent territories in which it has obtained the fewest votes, are considered to be merged into one territory. If necessary, a further merger is carried out of the two adjacent territories which then have fewest votes, and so on, until the party has as many territories as it has seats, and one candidate can be elected from each territory. In a non-merged territory which has more than one candidate, that candidate is elected who has most votes. In a merged territory, that candidate is elected who has most votes, in that one of the original component territories which had most votes.

If an MP dies or retires during the legislative term, then the seat is given to that candidate who obtained most votes, out of those candidates of the same party who stood in the same territory. If there was no other candidate of the same party standing in the territory, then one is elected from the top of the party's list of additional candidates.

These rules can best be illustrated by a hypothetical example. Figure 1 overleaf shows an extremely stylised map of 36 tracts, in which 36 MPs are to be elected. There are four parties, each creating a number of territories as follows: Labour (L) 14; Conservative (C) 12; Liberal Democrat (D) 9; Green (G) 2. Thus Labour creates eight territories each of three tracts, and six territories of two tracts, totalling fourteen territories, and it puts one candidate in each territory. For example, the Labour candidate L1 stands in a territory consisting of the three tracts on the top left. But in the event Labour gets enough votes to obtain only 13 seats, one less than its number of territories. Therefore the two adjacent territories in which it got fewest votes are merged, that is, the two territories on the bottom left consisting of the four tracts occupied by Labour candidates L12 and L13. L13 has more votes in his original territory than L12 has in his, and so L13 is elected to the merged territory.

The Conservative Party creates 12 territories, each of three tracts, and it puts one candidate in each territory. It gets enough votes to obtain 12 seats, and so it elects all its candidates.

The Liberal Democratic Party creates nine territories, each consisting of four tracts. In six of these territories it puts one candidate, and in three territories it puts two candidates. Since it gets nine seats, it elects one candidate in each of its nine territories, electing the candidate who got the greater number of votes, in each of the three territories where it has two candidates.

The Green Party creates two territories each of 18 tracts, and it puts two candidates in each territory. It gets two seats, and therefore it elects one candidate in each territory, that one who got more votes.

L1 C1 (D1) D2 G1 (G2)	L1 C1 (D1) D2 G1 (G2)	L1 C1 D3 (D4) G1 (G2)	L2 C2 D3 (D4) G1 (G2)	L2 C2 D5 G1 (G2)	L2 C2 D5 G1 (G2)
L3 C3 (D1) D2 G1 (G2)	L3 C3 (D1) D2 G1 (G2)	L3 C3 D3 (D4) G1 (G2)	L4 C4 D3 (D4) G1 (G2)	L4 C4 D5 G1 (G2)	L4 C4 D5 G1 (G2)
L5 C5 D6 G1 (G2)	L5 C5 D6 G1 (G2)	L5 C5 D7 G1 (G2)	L6 C6 D7 G1 (G2)	L6 C6 D8 (D9) G1 (G2)	L6 C6 D8 (D9) G1 (G2)
L7 C7 D6 G3 (G4)	L7 C7 D6 G3 (G4)	L7 C7 D7 G3 (G4)	L8 C8 D7 G3 (G4)	L8 C8 D8 (D9) G3 (G4)	L8 C8 D8 (D9) G3 (G4)
L9 C9 D10 G3 (G4)	L9 C9 D10 G3 (G4)	L10 C9 D11 G3 (G4)	L10 C10 D11 G3 (G4)	L11 C10 D12 G3 (G4)	L11 C10 D12 G3 (G4)
(L12) C11 D10 G3 (G4)	(L12) C11 D10 G3 (G4)	L13 C11 D11 G3 (G4)	L13 C12 D11 G3 (G4)	L14 C12 D12 G3 (G4)	L14 C12 D12 G3 (G4)

Whatever form of Proportional Representation is used for the Westminster elections, whether Territorial PR or some more conventional system, and even if the system is complemented by a relatively high representation threshold, it can be expected that there will be a substantial increase in party fragmentation, and a loss of seats by the larger parties to the smaller ones. Moreover, parties will tend to become more extreme, or at least to become more sectionalist, that is, tending to appeal, not to the whole nation, but to a limited section of it. The multi-party coalitions which this situation will require, are likely to be more unstable and ineffective, and perhaps beholden to some minor party which holds the balance of power.

This paper therefore proposes a few simple modifications to PR, which can be expected to greatly mitigate these problems, if not to overcome them. Referred to as Consensus PR, this modified system seeks to maintain three major parties and to discourage them from splitting, and gives each of them the incentive to remain or to become a one-nation party, responsive to each section of the electorate. It is argued that Consensus PR can thus be expected to achieve stable and effective government, responsive to and acceptable to each section of the electorate.

(1) It is a form of PR which gives positive incentives for the formation of three major parties, each with an average of at least 27% of seats.

(2) Hence two of these three major parties will be able to form a two-party coalition government with an adequate majority of seats. Thus the possible instability of multi-party coalitions is avoided, as also is the situation where a minor party holds the balance of power and so exerts excessive influence.

(3) These three major parties get their seats in proportion not only to their first-preferences, but also to their second-preference votes. This gives them the incentive to appeal more widely, to all sections of the electorate.

(4) Hence each major party, and any coalition government formed by any two of them, will tend to be moderate in policy, and responsive to all sections of the electorate.

(5) The minor parties receive up to 19% of seats, which are given in proportion to their first-preference votes. Since a two-party coalition government can always be formed by two of the major parties, there is no need to use a representation threshold, in order to reduce the number of small parties which get seats. Thus under Consensus PR, without a representation threshold, much better representation of minorities can be achieved, without incurring any risk to the effectiveness of government.

(6) Consensus PR is essentially a method of determining the number of seats of each party. Hence it must be grafted on to some other form of PR, which can allocate each party's seats to individual candidates, and allocate these candidates to geographical territories. Thus Consensus PR could be used in conjunction with some system such as Party List PR, Single Transferable Vote PR, the Additional Member System, or (in the opinion of the writer, most suitably) Territorial PR.

The ballot paper is in two parts (or there are two ballot papers). In the first part, the candidates are listed, according to the rules of whatever form of PR is being used to allocate seats to candidates, and the elector votes between them according to these rules. In the second part, the parties are listed, and the electors vote on them by putting the parties in order of preference, writing in "1" for a first preference, "2" for a second preference, and so on, for as many as they wish.

The three parties with most first-preference votes in the second part of the ballot paper, are designated as the three "major" parties, and the others as the "minor" parties. Each minor party gets a percentage of seats equal to its percentage of first preferences. Each of the three major parties gets a share of the remaining seats in proportion to the following: the number of voters who vote it the most-preferred out of the three major parties plus the number of voters who vote it the second-most-preferred out of the three major parties. If a person votes for only one of the major parties, they are counted as having given one vote to that party, and half of a second-preference vote to each of the other two major parties.

For example, suppose A, B and C are the three major parties, and X is a minor party. (For convenience, a person who votes A first, B second, C third, will be described as voting ABC.) A person who votes ABC, AB, XABC, XAB, AXBC, or AXB, is counted as having given one vote to A and one vote to B. A person who votes A, XA or AX, is counted as having given one vote to A, one-half of a vote to B, and one-half of a vote to C. Each major party then gets a share of the major-party seats in proportion to the number of these votes which it obtains.

Attention is drawn to the following characteristics of this method of allocating the major-party seats.

(1) A second-preference vote is worth just as much as a first-preference vote, in terms of seats. The rationale of this is that it gives the major parties a stronger incentive to avoid sectionalism, and to appeal to all segments of the electorate. This will be argued in more detail below.

(2) However a person votes for the major parties, whether they vote for one, two, or three of them, the person still contributes a total of two votes to the major parties, and not more than one vote to any one of them. It follows that there is no scope for a person to vote "strategically", in order to benefit their first-preferred party, by voting only for that party. Thus if they have a preference between the other two parties at all, it is to their advantage to express it.

(3) A person who votes first preference for a minor party, can still cast two votes for two of the major parties (or one to one and one-half to each of the other two). Thus the major parties have about as much incentive to appeal to the minor-party supporters, as they have to those who vote first-preference for a major party.

Another important feature of the system is that it gives an incentive for parties to merge, if it looks at all likely that the three major parties will get an average of less than 27% of seats. This incentive is provided as follows. If the average number of first-preference votes which the three major parties get, is less than 27%, then a percentage of seats is transferred, to those two major parties with most first preferences, from the other parties, both major and minor.

For example, if the average is 22%, then each of the two major parties with most first preferences receives 27 - 22 = 5% of extra seats. This total of 2 x 5 = 10% of seats is taken from the remaining parties, from each in proportion to its first preferences. Thus provided the two major parties with most first preferences do not get a less-than-average share of the initial allocation of major-party seats, they will get in total at least 2 x (22 + 5) = 54% of seats, an adequate majority for them to form a coalition government.

But if the third major party is expecting this, it will have an incentive, in order not to lose seats, to merge with one or more minor parties, so as either to form a party which has most first preferences or second-most first preferences (and so does not lose seats), or to bring the average major-party first preferences back up to something above 27% (so that no party loses seats). Thus the system tends to create three major parties with an average of at least 27% of first preferences and of seats. If at any time the major parties appear likely to drop below 27%, one or more of them will have the incentive to merge with smaller parties, so as to bring them back up to something above 27%.

How then will this Consensus PR system operate in practice, if used in the UK? Let us first consider the minor parties. At present, under Plurality, they get about 7% of the votes. But under any form of PR, they will greatly improve on this. If Territorial PR is used, without a representation threshold, or any other system of PR giving seats in one nation-wide constituency, they will increase their seats even more. Each minor party, however small, will get seats exactly in proportion to its first preferences over the whole country. Thus the number of minor parties will tend to increase, since it will be easy for new parties to get established and to grow, and there will be nothing to deter old ones from splitting. The total number of seats of the minor parties will also tend to increase over time.

But the seats of the minor parties will reach a ceiling, as the average number of seats left to the major parties is brought down close to the critical level of 27%. Then the smallest of the major parties will seek to merge with one or more minor parties, in order to avoid losing seats. On their part, the politicians of the minor parties are likely to be willing to merge with it, for two reasons. First, they also will be faced with the threat of loss of seats, which they will seek to avoid. Second, it may well be attractive to them to be able to join a potential government party. Thus the seats of the minor parties are unlikely ever to rise above this ceiling of 100 - (3 x 27) = 19%.

The long-term result of this will be that while new minor parties will more or less continually arise, other minor parties will disappear as they willingly merge themselves into the major parties, thus joining a potential government party. Thus the total of seats of the minor parties is likely to remain fairly constant, at not more than 19%, just as the total seats of the three major parties will remain constant at not less than 81%.

But what will be the effect of Consensus PR on the policy of the parties, as distinct from their seats? The major parties will get their seats in proportion to their first and second preferences. This will create a powerful incentive for a party to avoid any extremism or sectionalism, as can be illustrated by the following example.

Let us suppose that at the first Consensus PR election, Labour, the Liberal Democrats and the Conservatives respectively get 40, 20 and 30% of first preferences, and that the minor parties get 10%. Thus the minor parties will get 10% of seats. But how many seats a major party gets, will depend on how many voters make it their second-preferred major party. If let us say the Conservatives take an ultra-extreme and sectionalist position and so get no second preferences (that is, all they get is last preferences) from Labour, Liberal Democrat or minor-party first-preference supporters, then they will get only (30/200) x 90 = 13.5% of seats. But if on the other hand they take a moderate position and appeal widely to all sections, and so get an equal one-third share of the second preferences, they will get 30% of seats.

Thus an extremist or sectionalist major party could lose half of its seats, or more, by not getting second preferences. In terms of seats, second preferences are worth just as much as first preferences. In order to get one-third of the major-party seats, a major party must get either a first- or a second-preference vote from two-thirds of the voters. Each major party will thus have the incentive to appeal widely, to all sections of the electorate and to the supporters of all other parties, since there is no other way to get the required number of second preferences.

Thus all three major parties will be appealing to the same electors, that is, to all electors. The major parties can thus be expected to converge towards each other in policy. This convergence obviously makes for more stable government, whichever two of the major parties form the government coalition. Another factor contributing to stability of government, is the virtual certainty that there will be two of these major parties which in coalition have an adequate majority of at least 54% of seats. Consensus PR can thus be expected to provide stable and effective two-party coalition government, which is responsive to each section of the electorate.

But what type of policy will be followed by the minor parties? Unlike the three major parties, which will be broad in appeal and responsive to each section of electors, the minor parties will be niche parties, narrow in focus and divergent from each other. But in the context of this system, where their presence in the government will not be needed, this will create no problem. Indeed, the minor parties will serve a useful function in articulating minority interests and concerns, bringing them to the attention of the major parties and of the wider public. The major parties, which will be competing with each other for the lower-preference votes of these minorities, will therefore have a stronger incentive to respond to their concerns.

The conventional wisdom is that, with ordinary PR, there is a trade-off between effectiveness of government and the proper representation of minorities, this trade-off being manipulable by means of the size of the representation threshold. But under Consensus PR, where no threshold is needed, this trade-off disappears, and we get, if not the best, the better of both worlds, both greater effectiveness, and better representation.

Democracy Design Forum, Coles House, Buxhall, Stowmarket, Suffolk IP14 3EB, UK
Tel [int. 44] (0)1449 736 223
Fax [int. 44] (0)1449 612 274
e-mail: chapman@democdesignforum.co.uk

Democracy Design Forum

A Network for the Enhancement of Electoral and Constitutional Systems and Democratic Processes

TWO ELECTORAL SYSTEMS FOR WESTMINSTER

A submission to the Independent Commission on Voting

David Chapman

29 March 1998