Reform of the British electoral system has been much discussed in recent years. It is advocated by all centre parties—by the present Liberal Democrat Party, by its predecessors the Liberal and the Social Democrat Parties, and by the Green Party as well. The Labour Party, as a body, has maintained an aloof attitude, though it contains some strong advocates, as well as some strong opponents, of reform; but there have been clear indications that the Party might change its official stance on the matter. The clearest of these was the setting up in November 1990 of a Working Group, under the chairmanship of Professor Raymond Plant of Southampton University, to review the whole question. The Working Group has eighteen members, all Labour Party supporters, including six Labour mps, one of them Bryan Gould, two Labour members of the House of Lords, and one Labour mep. The present Report, issued in July 1991, is thus a good indication of current thinking on this issue within the Labour Party: though it is unfortunately not well known, it ought therefore to be widely studied.footnote1

It is an interim report, to be followed by one offering more definite conclusions. It aims to do no more than set out the nature of the problem, and the considerations in favour of competing solutions to it. It does not attempt to decide what system ought to be adopted, or even whether change is necessary at all, but only to set out the basis for such a decision: what questions we must answer in order to arrive at a decision whether we need a new system, and, if so, which. The Report has the great merit of making plain how complex the problem is; it rightly emphasizes that it is not a matter merely of devising the best mechanism to achieve an agreed end. To decide on an electoral system, you have first to decide what effect you want it to produce. This depends, in turn, on political judgements, of principle as well as of practical effect. In what sense should Parliament represent the nation? What function do we want it to fulfil? Can coalition governments be effective? Do we want public policy to change only gradually, as coalitions shift, or is it better that, at any time, government policy should be based on some clear ideology, to be reversed when that ideology fails of popular support? But, for all that it clearly conveys the breadth of the question, the Report suffers, as an analysis of fundamental principles, from a number of defects.

The topic is complex because, from any standpoint, there is not just one condition we should like an electoral system to fulfil, but several. Although, in my view, it fails to mention some of those conditions, the Plant Report strongly emphasizes this fact, and insists that, in consequence, any choice of system must involve a compromise. One such compromise it stresses is the concession made by most proponents of pr to the principle of local representation. The guiding principle of pr is that of proportionality, namely that the composition of Parliament, by political parties, should match as closely as possible the preferences of the electorate as between the parties. Maximal realization of this principle would require the entire country to be treated as a single constituency, as in Israel; but most advocates of pr are willing to temper the principle by some compromise with the requirement that all or at least some mps should represent a region or local constituency. The Report is certain that electoral reform in Britain should respect the principle of local representation, and inclined to believe that this should continue to be done by means of single-member constituencies rather than multi-member ones.

What it fails to do is to generalize this point, which arises even for those who do not adhere to the principle of proportionality. In a general election under any system providing for local representation, a voter is asked to help to determine two different things: his local representative; and the overall composition of Parliament. This creates difficulties both for the deviser of an electoral system and for the voter. For one devising an electoral system because, even when he is clear both how Parliament should be made up, given the preferences of the voters in the country at large, and which candidate or candidates should represent a constituency, given the preferences of the voters within it, a system designed to produce one outcome may well not produce the other. For the voter, because his preferences between the local candidates may not match his preferences between the national parties, or because he would like to vote tactically in the local election, but not to betray the party he most favours; he is then in a dilemma how to vote. The Additional Member System (ams) in use in Germany solves this problem, simply but brilliantly, by the device of giving each voter two votes, one for each of the two purposes. The Plant Report discusses the ams very sympathetically, and points out that either of its two components may be modified: but, just because it does not recognize the generality of the problem, it fails to give sufficient credit to the ams for this, its principal merit. Thanks to this idea, we need no longer worry whether the same mechanism will serve both ends, securing the right composition of Parliament and selecting the right constituency representatives. We do not have to decide in advance to use different mechanisms for the two purposes: the knowledge that we can do so is sufficient to remove a great part of our difficulty.

The selection of a voting procedure is problematic even for a fixed body, a committee, board or the like, making decisions independently of others; we might call this ‘the simple case’. The simple case is closely similar to the problem how one or more representatives ought to be chosen for a single constituency: electoral reform has to treat both of this and of the problem how to determine the composition of Parliament as a whole, problems treated together in the absence of the ams two-vote device. Politicians and journalists who discuss electoral reform usually fail to appreciate how complex even the simple case is: as a result, they concentrate exclusively on the problem peculiar to political elections, how the composition of Parliament should be determined. The Plant Report is not guiltless of this fault: despite the possibility of dissociating the two by means of the two-vote device, it tends to treat the method of electing constituency representatives solely in the light of its effect on the composition of Parliament, and to pay little attention to how fair a means it may be of choosing a local representative.

The fact that the Report mentions no more than glancingly the dilemma of a voter casting a single dual-purpose vote is part of a general failure to discuss voters’ dilemmas under different electoral systems. This involves an extraordinary omission: tactical voting, which is an attempt to resolve such a dilemma, is barely mentioned. The Report states (p. 20) that it will make little use of the mathematical theory of voting, a theory extensively developed in the last thirty years. This is a serious mistake, in my view; it is, on the face of it, improbable that a rigorous theory developed by systematic inquiry should be barren of results of practical relevance. The mathematical complexities can indeed be ignored: the fundamental concepts ought not to be.

One such concept that the Report does mention (p. 66), but then largely ignores, is that of a candidate who is a ‘Condorcet winner’; such a candidate is better called a ‘Condorcet top’, since he is not guaranteed to win under most electoral systems. Suppose that, in a series of ballots, the candidates were pitted against each other in pairs: in each ballot, one of the two candidates involved would win (barring ties, which would be rare). Assuming that the preferences of the voters remained the same, and that they always voted according to their true preferences, the series of ballots would show, for every two candidates, which of them was preferred by a majority to the other. A candidate who would win every ballot in the series is called a ‘Condorcet top’. The definition does not, of course, depend on such ballots actually being held, but relates solely to the voters’ preferences between the candidates, supposing them to be determinate.