Published online by Cambridge University Press: 27 January 2009
This article explores the properties of paradoxical voting situations in legislatures as illustrated by the decision-making process attending passage of the 1951 Electoral Reform Act in Fourth Republic France. First, conditions for demonstrating existence of a voting paradox for the case are developed and applied to data describing the voting behaviour of individual deputies. There follows an exploration of how a voting cycle over reform alternatives was overcome, focusing on the ways institutional rules and procedures structured the behaviour of parties and individual members and on the strategic manœuvring of political leaders. We conclude by relating characteristics of the case to general propositions found in the theoretical literature on the paradox of voting.
1 Arrow, Kenneth, Social Choice and Individual Values, 2nd edn (New Haven, Conn.: Yale University Press, 1963).Google Scholar
2 Black, Duncan, The Theory of Committees and Elections (Cambridge: Cambridge University Press, 1958).Google Scholar
3 Farquharson, Robin, Theory of Voting (New Haven, Conn.: Yale University Press, 1969).Google Scholar
4 It is fair to say that theoretical attention has focused largely on examination of the properties and implications of sophisticated voting and agenda manipulation in legislative settings. Perhaps this stems from the fact that sophisticated voting constitutes a rational strategy available to self-interested voters. As such, analyses become theoretically enriched in so far as sophisticated voting implies an interaction of individual preferences and political structure (institutional rules), opening an avenue for modelling the deceit and manipulation commonly associated with political decision making. The ‘natural’ occurrence of a voting paradox appears less interesting, theoretically, since it does not arise as the result of a strategy but rather as an ‘accident’ stemming from particular preference distributions across actors. Since social choices from a naturally occurring paradox of voting are not due to strategic behaviour, they are ‘unmotivated’ from the vantage point of theoretical explanation, and imply simply that the political process can generate arbitrary results. Given the general disciplinary interest in deriving theoretical explanations of important political phenomena, it is not surprising that the few attempts to demonstrate empirically the existence of a paradox of voting have focused on cases involving an assumption of sophisticated behaviour. See Riker, William, ‘Arrow's Paradox and Some Examples of the Paradox of Voting’, in Claunch, J. M., ed., Mathematical Applications in Political Science, I (Dallas, Tex.: Arnold Foundation, Southern Methodist University, 1969), pp. 41–60Google Scholar; and Riker, William, Liberalism Against Populism: A Confrontation Between the Theory of Democracy and the Theory of Social Choice (San Francisco: Freeman Press, 1982).Google Scholar
5 Shepsle, Kenneth and Weingast, Barry, ‘Uncovered Sets and Sophisticated Voting with Implications for Agenda Institutions’, American Journal of Political Science, 28 (1984), 49–75.CrossRefGoogle Scholar
6 A backward-moving agenda process is an ordered series of proposals, consisting of a motion to alter the status quo (a counter bill) and subsequent amendments to that motion. At some point, further amendment is not allowed and voting begins, considering each amendment in reverse order to its introduction. Under forward-moving agenda, amendments to a motion are voted as they are received, successively redefining the content of the counter bill. The final vote under both agenda processes implicitly pairs the counter bill, as amended, with the status quo, and it is voted up or down. Since a counter bill is successively amended under forward-moving agenda, actors lack substantial control over the timing and content of amendments that are introduced, making the design of a strategy aimed at determining the (distant) outcome of the final vote very difficult. The strictly backward ordering of votes over amendments makes strategic control of the final vote a more tractable matter. See Wilson, Rick and Pearson, Anne, ‘Evidence of Sophisticated Voting in a Committee Setting: Theory and Experiments’, Quantity and Quality, 21 (1987), 1–19.CrossRefGoogle Scholar
7 Parliamentary procedure in the Fourth Republic provided that any bill that is introduced may be successively altered without formal recourse to its original content (the bill is altered or amended as it goes). Thus, the bill's content changes with each succeeding amendment until some arbitrary point is reached when a final (up or down) vote is taken. This procedure may be described as follows (see Lidderdale, D. W. S., The Parliament of France (London: The Hansard Society, 1951), pp. 175–6)Google Scholar: (a) presentation of a bill and immediate referral to an appropriate committee; (b) committee consideration (including redrafting), preparation of a report, and presentation to parliament; (c) general debate on committee report, culminating in a vote to consider clauses; (d) debate on clauses including moving and voting amendments; (e) final debate and voting on the bill as a whole (first reading) as it has emerged from the preceding steps.
8 To illustrate, consider two sets of orderings over three alternatives for three voters. For Voter 1, alternative a is preferred to b and b is preferred to c. Transitivity establishes that aPc: thus, aPbPc. For Voter 2, bPaPc, and for Voter 3, cPbPa. When a is paired with b, b is preferred by a majority (Voters 2 and 3). When c is paired with b, b is also preferred by a majority (Voters 1 and 2). Since alternative b is preferred by a majority to all other alternatives, it is the Condorcet Winner for this situation. However, if all voters only voted for their most preferred alternative and against all others, a voting cycle (serial defeat) would be the outcome. Consider next the orderings: Voter 1, aPbPc, Voter 2, bPcPa, and Voter 3, cPaPb. In this case there is no alternative that is preferred by a majority of the voters to some other alternative when all pairs of alternatives are considered. It is in this circumstance that a paradox of voting exists.
9 MacRae, Duncan, French Representation Study, 1946–1958 (Ann Arbor, Mich.: ICPSR, 0052, University of Michigan, 1961)Google Scholar; MacRae, Duncan, ‘Intraparty Division and Cabinet Coalitions in the Fourth Republic’, Comparative Studies in Society and History, 5 (1963), 164–211CrossRefGoogle Scholar; MacRae, Duncan, Parliament, Parties and Society in France: 1946–1958 (New York: St Martin's Press, 1967).Google Scholar
10 MacRae's source for roll-call data was the Journal Officiel, the authoritative source of parliamentary proceedings for France. Apart from this source, there are also generally other reports of roll-call voting available, usually presenting the breakdown of the vote by party support (not individual deputy). Cross-checking MacRae's data with other sources (such as Le Monde, Keesing's Contemporary Archives, and tables appearing in standard texts), some reporting discrepancies may be noted. Whether these result from coding errors in the MacRae dataset or are attributable to other sources is difficult to ascertain without recreating MacRae's dataset from scratch. In any event, the discrepancies are not large in most cases and are most commonly observed when aggregated for reporting purposes to the level of party. Here, inconsistent grouping criteria across sources may be an additional source of error. This is at least minimized in those parts of our analysis that assess behaviour at the level of the individual deputy.
11 Williams, Philip, Crisis and Compromise: Politics in the Fourth Republic (New York: Doubleday, Anchor Books, 1966), p. 323.Google Scholar
12 Goguel, François, France Under the Fourth Republic (Ithaca, NY: Cornell University Press, 1952), p. 68.Google Scholar
13 Williams, , Crisis and Compromise, p. 324.Google Scholar The qualification inserted into the Williams quotation recognizes a shift of position from proportional representation to a majority voting system introduced by the MRP leadership at the party's conference in May 1950. In spite of rank-and-file resistance, the MRP parliamentary leaders over the next few months crafted a one-ballot majority proposal that included list alliances, while simultaneously expressing opposition to any two-ballot majority system (the traditional position of the Radical Socialists).
14 Goguel, , France Under the Fourth Republic, p. 70Google Scholar, states that Giacobbi adapted his proposal from a bill that had been introduced some years earlier by René Coty (Independent Republican, then Senator and later to succeed V. Auriol as president of the Fourth Republic). In addition to the main provisions of the bill, cited above, it also forbade electoral alliances on either ballot, permitted preferential voting and panachage, and instituted a procedure for holding by-elections.
15 Goguel, , France Under the Fourth Republic, p. 71.Google Scholar
16 In our analysis we make no analytic distinction between the two positive proposals to adopt a proportional representation law that were offered and the 1946 law that would govern the 1951 elections if no reform proposal was adopted by the Assembly. These were viewed by all parties as alternative ways to accomplish the same thing, a 1951 election under PR rules. In view of the fact that much of the electorate and many MRP and SFIO deputies and cadres had a principled attachment to PR, the PCF and its affiliates sought to exploit the determination of the centre majority and new right parties to abandon PR by characterizing their behaviour as cynical political opportunism that was contemptuous of democracy. The positive proposals for adopting a PR electoral law, then, are plausibly regarded as ways to expose the opportunism of the centre majority and new right parties by forcing them on the record in opposition to PR.
17 The parties of the centre majority viewed joint lists as critical to the success of their candidates when running in constituencies heavily populated with Gaullist or Communist supporters. Since neither the Gaullists nor Communists expected to benefit from list alliances (if for no other reason than a lack of likely allies), they favoured each party going alone to capitalize on the overall size of their respective constituency electorates.
18 In France, as elsewhere, a vote of confidence puts the existence of a government at risk. It incurs an obligation to resign if an absolute majority opposes its motion in Assembly voting. Failing such an outcome, the government is under no obligation to resign. In the present case, not only was the government not defeated, it actually commanded a relative majority in favour of its motion. However, the abstention of the MRP, a major government partner, implied a lack of confidence such that Pleven opened discussions with the president of the Republic over the matter of his resignation. President Auriol resisted this move on the ground that the government had not been defeated, and Pleven agreed to continue the process of seeking agreement on the issue.
19 It is comforting to note that when an actor's ‘true’ and ‘real’ preference orderings diverge, the ‘true’ ordering has no behavioural consequence, and thus contradicts the expectation of the rational decision model that behaviour follows directly from actor preferences.
20 It is for this reason that where some alternative is a Condorcet Winner, the expectation is that it will eventually be supported by some actors who prefer another alternative to it.
21 Quattrone, G. and Tversky, Amos, ‘Contrasting Rational and Psychological Analysis of Political Choice’, American Political Science Review, 82 (1988), 719–32.CrossRefGoogle Scholar
22 Black, , The Theory of Committees and Elections, pp. 4–11.Google Scholar
23 In Appendix 2 we speculate that the 116 MRP deputies in Column 2 might reasonably be classified as having the preference order bPaPc rather than the order b and not (a, c) that we attribute to them. Considering this possibility, the results of the analysis just presented will change the outcome for only one of the paired comparisons, that being with respect to the pair a, c where we presume that a is the proposed alternative. Here, the number of deputies that prefer a to c will increase by 116 from 207 to 323, while the number preferring c to a will decrease a like amount from 409 to 293. This change reverses the majority outcome for this pair with the result that proportional representation is preferred to a two-ballot majority system under both circumstances of voting. This change does not, however, overturn the general conclusion of the existence of a voting paradox. Rather, the result is that a one-ballot system beats proportional representation, proportional representation beats a two-ballot system, and a two-ballot system beats a one-ballot system whenever a one-ballot system is the current proposal.
24 Siegfried, André, L'Année Politique: 1951 (Paris: Presses Universitaires de France, 1951), p. 66.Google Scholar
25 Le Monde, 15 March, 1951, p. 1.
26 This is noted both by Neumann, Robert, ‘The Struggle for Electoral Reform in France’, American Political Science Review, 45 (1951), 741–55CrossRefGoogle Scholar; and Goguel, , France Under the Fourth Republic. The latter author offers the following interpretation (pp. 73–4)Google Scholar: ‘It required all the political skill of Henri Queuille, who returned as Premier, to persuade the National Assembly to pass the bill. He succeeded in doing this because the attachment of his Radical–Socialist colleagues to the single-member system with two ballots, which was ardently demanded at all their congresses, really represented the sentiments of the active party workers more than those of the party leaders … Everything considered, however, the majority system with alliances and with one ballot protected their interests almost as completely as would a list system with two ballots. That is why they eventually supported it.’
27 Lidderdale, , The Parliament of France, pp. 48–9.Google Scholar
28 Quoted from Siegfried, , L'Année Politique: 1951, p. 97.Google Scholar
29 Quoted from Le Monde, 29 and 30 04 1951, p. 4.Google Scholar
30 Siegfried, , L'Année Politique: 1951, p. 98.Google Scholar
31 Le Monde, 29 and 30 04 1951, p. 4.Google Scholar
32 Siegfried, , L'Année Politique: 1951, p. 98.Google Scholar
33 Indeed, Table 2 discloses that three UDSR and no Radicals held PR as a second preference, while, given the prior electoral system commitments of their party, the 116 MRP deputies listed in Column 2 may have held PR as their second preference. In addition, descriptions of the reform process indicate that the MRP leadership felt abandoning PR in favour of a (one-ballot) majoritarian system was a significant sacrifice of principle, and having taken that step, they were not prepared to go further (as far as a two-ballot system.)
34 Riker, William, ‘Political Theory and the Art of Heresthetics’, in Finifter, A., ed., Political Science: The State of the Discipline (Washington, DC: American Political Science Association, 1983), pp. 47–68Google Scholar; and Riker, William, The Art of Political Manipulation (New Haven, Conn.: Yale University Press, 1986).Google Scholar
35 The decision by MRP in the first stage to abstain on the vote of confidence might also be considered such a manœuvre, although in this case existing conditions were retained rather than redefined.
36 Riker, William, ‘Implications from the Disequilibrium of Majority Rule for the Study of Institutions’, American Political Science Review, 74 (1980), 432–46CrossRefGoogle Scholar; Black, , The Theory of Committees and ElectionsGoogle Scholar; and Arrow, Social Choice and Individual Values.
37 McKelvey, Richard, ‘Intransitivities in Multidimensional Voting Models and Some Implications for Agenda Control’, Journal of Economic Theory, 12 (1976), 472–82CrossRefGoogle Scholar; and Schofield, Norman, ‘Instability of Simple Dynamic Games’, Review of Economic Studies, 45 (1978), 575–94.CrossRefGoogle Scholar
38 This assertion is contradicted when a ‘final’ vote is taken, pairing some positive proposal for change with the status quo. As the legislative history of this issue will demonstrate, however, it was quite difficult for deputies to know which of the votes taken would determine the issue. At several junctures in the voting, a decisive outcome seemed to have occurred, only to be subsequently overturned and the process continued. In the end, final disposition of the issue appears to have resulted from the exhaustion of the deputies.
39 These generalizations about voting patterns do not refer to those deputies who were members of the cabinet. Prior to moving confidence in consideration of its own bill, the cabinet was officially neutral between one-ballot and two-ballot systems and abstained en bloc on all relevant votes while consistently voting against PR. Subsequent to this vote of confidence, upon which the MRP abstained, cabinet members individually cast positive (for and against) ballots indicating their relative preferences over the various alternatives.