January 16, 2003

Chapter 6

Methods of Electing Legislators

Legislators are ordinarily chosen through elections in voting districts. The methods of electing them vary greatly among districts. Both the presidential and parliamentary systems typically have a legislature whose representatives are elected from different voting districts. The purpose of this chapter is to discuss various rules for choosing these legislators. The members of a collective must make two basic decisions: (1) how many legislators to have and (2) how to elect them. We shall not have much to say about the number of legislators. It is perhaps enough to note that there does not appear to be any magic, optimum But there are limits. At one extreme, members of a collective would want to elect enough representatives to discourage conspiracies by the elected legislators that might increase the risk of power abuse. At the other extreme, a large number of legislators would be costly to maintain. Within these limits there is a lot of room for choice.

     The main concern of this chapter is the different methods that members of a collective may use to elect representatives. We shall explore three characteristics of such methods. The first, which is discussed in part one, is the number of representatives to be elected. The second concerns eligibility rules for becoming a candidate for the legislature. Should everyone be eligible or should candidates be required to meet some minimum qualifications in order to get on the ballot? Part two deals with these questions. The third characteristic concerns how to account for voter preferences in the actual tabulation of votes. Should each person be allowed only one vote or more than one? And if he is allowed more than one, how should his votes be prioritized? What weights should be assigned? We discuss these issues in part three.

     The choice of election methods should take account of the potential waste of resources. Part four discusses the potential waste due to strategies adopted by political parties to give them an advantage over their rivals. We discuss two sources of waste: gerrymandering and strategic action by political parties in multi-representative district elections.

     Among the democracies of the modern world, there are many variations in the number of legislative bodies. We have noted that although most democracies have a single legislature, the U.S. has two legislative houses. In addition, there may be a number of state, or provincial, legislatures as well as county, city and other local ones. In order to understand the problems of choosing a system of representation at the most elementary level, we must begin with a simple case. So we assume that the framers of the constitution have decided on only one house of the legislature. Furthermore, we disregard provincial, state, city, and local legislatures. Thus, we narrow our analysis to a national legislature with a single house. We further assume that the members of the collective have already decided on the total number of representatives they want in the legislature.

1. SYSTEMS OF REPRESENTATION

     We can distinguish between three systems of representation: (1) the single-representative-district system, (2) the multi-representative-district system, and (3) the single nationwide constituency.

To understand the first, let us suppose that members of the collective decide that they want a 200-person legislature. Then they would divide the nation into two hundred districts of roughly equal populations. The voters in each district would then elect one legislator. This is the single-representative district system. In the multi-representative district system, each district elects more than one representative. For example, members of the collective might divide the nation into ten districts, each of which is charged to elect 20 legislators.

Some democracies choose all of their legislature from a single nationwide constituency. Examples are the Netherlands and Israel. It is as though the nation is itself a single, multi-representative district. Legislators chosen in this way do not represent voters in a specific district. The framers of a constitution expect them to represent all districts taken together. Others nations choose part of their legislature in this way. In Taiwan, for example, approximately one-fifth of the legislature is elected in this way. In Chapter Four we showed why it would not be a good idea to have all the legislators chosen from a nationwide constituency. It would not make the best use of voter's existing information about candidates and it would not encourage them to acquire more information. The elected agents would be less likely to reflect voter preferences and the problem of monitoring legislators would be greater. However, there is also an advantage to the nationwide constituency. It is that legislators elected in this way have a lesser incentive to advocate policies that benefit the residents of a particular voting district at the greater expense of residents of other districts. Given the pros and cons, we cannot be certain which of these three systems will be best in a particular given situation.

2. ELECTION COMMISSIONS AND

CANDIDACY REQUIREMENTS

     Before members of a voting district can have an election, they must first select candidates. At one extreme, we can imagine a system in which a voter could vote for any person in the collective. Under this system, if you went to vote, you would be asked to select the name or names of the person(s) who you wanted to be a legislator(s). You could write anyone's name. Other voters could do the same. No democracy currently uses such a system to our knowledge. For the more technologically advanced countries, the reason may be that in the past, such a system would have been costly and also subject to many errors; since it would have been necessary to hire a large number of human vote counters. In the modern age of computers, many nations have the ability to use this method at relatively low cost. So perhaps we shall see it in the future.

The more common method is for candidates to register with an election commission. The method of selecting members of the election commission and the rules it must follow are either specified in the constitution or determined at the first legislative session. The commission is assigned the task of deciding whether candidates have met certain qualifications. Some qualifications may be simple and non-controversial, such as a requirement that the person be a citizen who has reached a certain age and that he announce his candidacy before a certain date. More dubious restrictions are rules that require a candidate to obtain the endorsement of a minimum number or percent of the population, to post a cash bond, or to be nominated by a political party. There are two possible economic justifications for restricting the number of candidates. First, more names require higher costs of administering an election. More importantly, if there are too many names, the cost to voters of finding their preferred candidate on a list may be high. Thus, charging a fee or imposing some other cost on the candidate is a means of reducing the negative external effects of crowding on the ballot. As before, however, in the modern age, the justifications seem weak.

     Eligibility rules may give an advantage to some individuals in becoming candidates. Fees and cash bonds may exclude popular candidates who do not have the backing of wealthy individuals or business groups. The requirement of being affiliated with a political party excludes challenges from candidates who do not wish to be aligned with an existing party, unless new political parties are relatively easy to form.

3. ACCOUNTING FOR VOTERS' PREFERENCES

     Given that a method for selecting candidates is decided upon, the framers must devise a means of registering voters' preferences. In this subsection we will consider three methods. Each one has some advantages and some disadvantages.

One-Voter-One-Vote

     One method is to allow each voter only one vote for each seat in the legislature that is allocated to the voting district. For example, if a district has been allocated five seats (i.e., if it has five representatives to the legislature), a voter could vote for only five names. Suppose that you are a voter. At the time when you voted, you would be presented with a list of, say, twenty-five candidates from which you could choose only five. In a single-member district, each voter could vote for only one name. Let us call this the one-voter-one-vote method. This is the simplest method.

     The one-voter-one-vote method can lead to an outcome that is misleading. This is easiest to see if we assume a single-representative district system. Suppose that there are five candidates for a single position as representative of the district. Each voter can only vote for one candidate. Assume that voter preferences are as follows. If each voter votes for his true first choice, four of the candidates would each receive approximately twenty-five per cent of the vote. The fifth candidate would receive no votes. But now consider the voters' second choice. Assume that every voter prefers the fifth candidate as her second choice. Given these preferences, a good case can be made that even though the fifth candidate would receive no first-place votes, she is most preferred by voters. The one-voter-one-vote method is misleading because it does not indicate voters' preferences for candidates who are not chosen first. It does not allow a candidate who is not the first choice to be elected, even though she may be the most popular of the candidates.

     We can apply similar reasoning to the multi-representative district system. Suppose, as in our first example, that five representatives are elected from a district. There may be a popular candidate who does not rank among the first five on the scales of most voters. Such a candidate would not be elected even though we might agree that she is preferred by voters taken as a whole to some or all of the five candidates who receive the highest number of first-place votes. We might say that the one-voter-one-vote method has the potential of discriminating against candidates who, although they are popular second choices, do not rank first on most voters' preference ranking.

One-Voter-Multiple-Vote

A second method of selecting candidates is to allow the voter to vote for several candidates for the same seat. We can call this the one-voter-multiple-vote method. If this method is used, then a decision must be made on how many candidates a voter can vote for and on how the ballots will be tabulated by the election committee. Let us again refer to the single-representative district system. Assume that there are five candidates. Also assume that the election rule states that each voter can vote for three. Finally, assume that the committee decides to assign an equal weight to each vote. In this case, each of the three candidates for whom you vote would receive one point from you, while the other two candidates would receive no points. The candidate with the highest number of points from all voters would win. We can call this the equal-point method of tabulating ballots.

     Another way of tabulating ballots in the one-voter-multiple-vote method is to assign weights to voters' choices based on the voter's priority ranking. According to this method, the election committee might assign three points to your first choice, two points to your second choice, and one point to your third choice. In tabulating the ballots, they would count the number of points received by each candidate. As before, the candidate with the highest number of points would win. We can call this the weighted-point method of tabulating ballots. If we used this method, we might prefer that each voter would vote for all five candidates. In this event, we would assign four points to a voter's first choice, three points to her second choice, and so on.

Still another way of tabulating ballots can be called the points-assigned-by-voter method. According to this method, the election committee gives each voter a number of points which he is asked to divide up among the candidates. For example, in a five candidate single-representative district election, you may be given one hundred points. You must then assign your points to candidates in such a way that the total number does not exceed one hundred. If you like two candidates equally but dislike the others, you could assign fifty points to each one. Or if you like one of the two better, you might assign ninety points to her and ten points to the other. A variation of this method might include negative points. For example, you might be allowed to assign a maximum of, say, one hundred negative points to a candidate who you do not want to win.

Approval Voting

     A third method of selecting candidates which has become especially popular among Public Choice theorists in recent years is called approval voting. Under approval voting, you would choose among the candidates on the basis of whether you approve of them. In the case of five candidates in a single-representative district, you may approve of two candidates and disapprove of three. Then you would vote for the two candidates. The election committee would tabulate by assigning one point to each of the two. If you approved of only one candidate, you would vote for only the one. You could vote for all the candidates of which you approved, and each one would receive one point.

Pluralities and Majorities

     A simple majority refers to more than fifty per cent. Other majorities include a two-third's majority and a three-fourths majority. As pointed out in Chapter Four, a plurality refers to the largest number of votes, regardless of whether it is a majority. Consider a single-representative district system with a one-voter-one-vote method of tabulating ballots. Some election rules specify that a candidate can win an election by only receiving a plurality. Other rules specify that to be a winner, a candidate must receive a simple majority. If there are only two candidates, it is almost certain that one candidate will win a simple majority of the votes. Otherwise, there would be a tie. But in multi-candidate elections, the candidate with the most votes may not win a simple majority. If not, a nation that requires a candidate to gain a majority in order to win the election must do one of two things. First it can conduct at least one additional election -- one that excludes candidates who earn the lowest number of votes in the first election. We call such an election a runoff election. Second, it can use a vote tabulating system that permits it to make decisions on the basis of a hypothetical additional election. Consider each in turn.

A Runoff Election

     A nation may use a single runoff election or a series of runoffs. If a single runoff is used, the top two vote getters are pitted against each other in the runoff. If a series is used, the bottom vote-getter is eliminated and a second election is held among the remaining candidates. In the second election, if no one wins a majority, another candidate is removed and a third election is held. The process continues until one candidate receives a majority of votes.

A Hypothetical Additional Election

     Instead of having actual elections, a nation may use hypothetical elections. It is feasible to do this when the nation uses a one-voter-multiple-vote system. Assume that after the votes are counted and the points assigned, no candidate wins a simple majority. Then the committee must eliminate one or more of the low vote-getters. It then tabulates the votes a second time assuming that the low vote-getter(s) is not a participant. In other words, it assigns points in a proportional way under the assumption that the low vote getter(s) was not part of the first election. If there is still no majority winner, the process is repeated by eliminating one or more additional candidates from the running. The process is continued until one of the candidates wins a majority.

4. STRATEGIC ACTION BY POLITICAL PARTIES

Gerrymandering

Let us suppose that the framers of a constitution chose geographical voting districts with roughly the same number of voters in each. Assume for simplicity that each is a single-representative district. The first elections are held and a representative is chosen from each district. Every few years, new elections are held; either incumbents are reelected or challengers succeed in unseating the incumbents. As the years pass, the populations in some districts would increase while the populations in other districts might decline or increase less rapidly. Eventually, there might be very large differences between the populations in different districts. If community members wanted to maintain the original goal of roughly equal representation, they would have to redraw the voting district boundaries.

     The particular method they choose to redraw the boundaries may effect the outcome of subsequent elections. The main reason is that the popularity and efficiency of political parties in obtaining votes for their candidates differ in different districts. Take a simple example. Suppose that there are three hundred voters divided into three single-representative districts. Assume that there are two parties, conservatives and liberals. Assume further that for the whole country, the conservatives enjoy a 60%-40% advantage over the liberals and that there are no independent voters. One of the districts has become too large and another has become too small. So redrawing the boundaries becomes an issue. The question is how the boundaries should be redrawn.

     Each party would want the boundaries redrawn so that it could win as many seats as possible. So the conservatives would want to recommend a plan that would give them a majority in each district. Plan A in table 1 is a possible redistricting plan that would accomplish this goal. Such a plan assures that conservatives win all three seats. The liberals, however, would want to recommend a plan like plan B, which would assure that they would win two seats.

     Gerrymandering refers to attempts by a party to cause the boundary to be drawn in its favor. The term is a combination of "Gerry," which is the surname of an early 19th century governor of Massachusetts, and "salamander," which describes the geographical shape of one of the voting districts after his party was successful in redrawing a voting district boundary in its favor. Referring to our example, suppose that district boundaries were decided by a simple majority vote in the legislature. If the conservatives had a majority they would choose a plan like A; if the liberals were in power, they would choose a plan like B. Over time, as the power shifted and populations shifted from one district to another, the advantages of redrawing the geographical boundaries would change. But in each case, the party in power would have an incentive to make the boundaries in such a way that it maximized the number of seats it would win in the election.

     It is easy to see how a system that permitted the easy redrawing of boundaries by the legislature would lead to a waste of resources. Whenever a new party came to power, it would try to pass a law to redraw the voting district boundaries. A means of avoiding this waste is to establish a formula in the constitution. Or, if possible, a non-partisan agency like an independent judicial branch of government could be directed to make boundary decisions when appropriate. In the latter case, the redrawing decision should be based on general principles, including the principle that the resources used by political parties to press for a change be minimized.

     It is worth noting that even if we stopped gerrymandering, voters themselves could achieve the same effect by deciding to move from one district to another. However, the cost of moving is ordinarily not low and the benefit to any single voter is ordinarily not high. So we would ordinarily not expect to observe much of this activity.

The Party List System in a Multi-Representative District

      Every democracy has political parties. Because of this, candidates for election are ordinarily permitted to show their party affiliations. In other words, the ballot lists not only the candidate's name but also her party. This is especially useful for voters who know nothing about the candidates, since they may have a definite preference for the candidate's party or for principles that they believe the party stands for.

     An election system that is used in many nations that have multi-representative districts is the party list system. According to this system, the leaders of political parties choose the candidates. In the simplest case, the leaders also choose the order of the candidates. To see how a simple party list system works, suppose that there are three parties: conservative, liberal, and progressive. Assume that voters from the district want to elect ten representatives. Then each party would submit a list of ten candidates. At the voting booth, voters would look over the names on the party lists and then vote for one of the three parties. They would vote only for the party, not for the candidates. After the votes were counted, seats would be allocated according to the proportion received by each party. For example, suppose that there are 1,000 voters and that 500 vote for the conservative party, 300 for the liberal party, and 200 for the progressive party. Then the top five candidates on the conservative party list would be elected, the top three on the liberal party list would be elected, and the top two on the progressive party list would be elected. Independent candidates could run but each candidate would be like a political party. To be elected, an independent would have to poll at least ten per cent of the vote.

The Party List System as a Means of Reducing Strategic Action

     The party list system is one means of solving an intractable problem of wasteful strategic action. To see why, assume that the multi-representative voting district does not use a party list system and that each voter can only vote for one candidate. Suppose that one candidate in the conservative party is so popular that 90 per cent of the conservative party members would vote for him. Assume that the rest of the conservative party vote is evenly split among the rest of the candidates. Finally assume that in the other two parties, the voting is evenly divided among all the candidates. Table 2 shows the candidates who would be chosen, given the assumptions about party membership in the last paragraph. Candidates from the conservative party would win one seat and candidates from the liberal party would win nine. The progressive party would win no seats.

     Obviously, the conservative party would not be satisfied with this outcome. If the party leaders anticipated such a result, they would advise their voters to change their votes. For example, they might first assign numbers to the candidates. Then they would instruct party members to vote for the candidate whose number matches the last digit in the voter's phone number or identity card. If voters followed their orders and if the leaders of other parties did nothing, then each candidate in the conservative party would receive approximately 50 votes. The conservative party would win all the seats.

     Of course the other parties would not be satisfied with this outcome. If they knew that the conservative party leaders had advised their members to divide their votes evenly, they would advise their own members to change their votes. Consider the following strategy. Leaders of the liberal party could advise their members to only vote for the top five candidates and to divide their votes equally among them. Assume that leaders of the progressive party give similar advice, telling voters to vote for only the top three candidates. Then the outcome would be as shown in table 3. The conservative party would win only two seats, the liberal party five seats, and the progressive party three seats.

     We could imagine a number of similar examples. The basic point is that in the absence of a party list system, there is no way to avoid the incentive of political party leaders to act strategically by advising voters to vote for candidates for whom they otherwise would not vote. In an effort to get the highest number of its candidates elected, political parties would waste resources trying to predict the strategies of the other parties and devising appropriate counter-strategies. Using a party list system avoids this waste.

     It is sometimes argued against the party list system that because the ranking of candidates is made by party leaders, voters have less choice than they would have if they could vote directly for the candidates. This problem could be avoided by having primary elections inside the party. Primary elections would help party leaders determine party members' orders of preference. But primaries are costly. A more important point is that primaries are not usually necessary because party leaders have a strong incentive to select popular candidates in order to appeal to voters who have only a weak or no preference for a particular party. If party leaders choose unpopular candidates, the party's popularity would decline and voters would shift to a different party or to an independent candidate. It is important to remember that in a competitive political system, the main goal of party leaders is to win elections.

Questions for Chapter 6

1.  Define gerrymandering, party list system.

2.  Suppose that the framers of a constitution wanted one hundred legislators. Explain how they could create a single-representative district system of representation. Then explain how they could create a multi-representative district system.

3.  Suppose that the framers of a constitution wanted one hundred legislators. Tell the advantage and disadvantage of electing all legislators from a single nationwide constituency, as opposed to dividing the nation into voting districts.

4.  Show that if the voters of a district use the one-voter-one-vote method of electing representatives, the result may be that the most popular candidate is rejected. Do not use the same example as the one in the text.

5.  Distinguish between the one-voter-one-vote method and the one-voter-multiple-vote methods of electing representatives.

6.  If the one-voter-one-vote method of electing legislators is chosen, a decision must be made by the election committee on how many candidates a voter can vote for and on how the ballots will be tabulated. Distinguish between the following three methods of tabulating ballots: the equal-method, the weighted-point method, and the points-assigned-by-voter method. (Hint: you can give an example of each.)

7.  Describe approval voting.

8.  Under what conditions would a runoff election be necessary in order to choose a representative legislator from a district. Assume in your answer that the government uses a single-representative district system.

9.  Suppose that there are several candidates in a single-representative district system. Assume further that the system uses a one-voter-multiple-vote system of electing the legislator from the district. An election is held but no candidate wins a majority of the voting points. Explain how hypothetical additional elections could be held to assure that a candidate wins a majority.

10.   Explain why gerrymandering leads to a waste of resources.

11.   How can gerrymandering be minimized?

12.   Describe the party list system. In your description use and example that is different from the one in the text.

13.   The party list system is a means of enabling voters to select among candidates in a multi-representative district. Describe the problem that the system helps to solve. In your description use an example that is different from the text.

14.   It is sometimes argued against the party list system that because the ranking of candidates is made by party leaders, voters have less choice than they would have if they could vote directly for the candidates. Do you agree? Why?

Gunning’s Address
J. Patrick Gunning
Professor of Economics/ College of Business
Feng Chia University
100 Wenhwa Rd, Taichung
Taiwan, R.O.C.
Please send feedback

Email: gunning@fcu.edu.tw

Go to Pat Gunning's Pages

Mirror Site of Pat Gunning's Pages