There’s Room in the Middle: Duverger, Downs, and the Liberal Party

In the run up to the 2015 election there is a debate over whether the Liberals or NDP offer the best opportunity for progressives to replace the Conservative government. Many anti-Conservative voters are willing to back whichever of the two parties has the best chance of winning, which could mean that the weaker of the Liberals and NDP loses large amounts of support to strategic voting. Through most of Justin Trudeau’s time as Liberal leader the Liberals have looked like the stronger of the two non-Conservative parties, but recent polls show the NDP ahead of them and in a very close race with the Conservatives. This raises questions as to what will become of the Liberal party if they finish third in the 2015 election. My colleague, David Moscrop, recently published this piece questioning whether the Liberals matter are still needed.

The increasingly left-right dynamics of Canadian politics mean that the Liberal party’s days as Canada’s “naturally governing party” are likely done, but that does not mean that the Liberals no longer have a place in Canadian politics. Not since the 1920s has Canada been a two party system, and small parties have an important place in Canadian political history. The threat of losing support to the CCF pushed Mckenzie King to support a more expansive welfare state in the mid-1940s, federal funding for medicare was introduced by a Liberal minority government that was reliant on NDP support, and the strength of the Bloc Quebecois in Quebec was in part responsible for the strong of minority governments that Canadians saw through the 2000s. The Liberal party should not be written off as unimportant should it finish third for a second consecutive time in 2015.

The Liberals and the NDP appear to be fighting what is known in political science as “Duvereger’s Law.” This is the argument that first past the post electoral systems, like the one used in Canada, should produce two-party systems. Because only one party can win a given district, it only makes sense for parties that finish first or second (other parties tend to have little chance of winning a district) to run a serious electoral campaign in any particular district. Duverger expects voters to make strategic choices, opting for one of the two parties with the best chance of winning. To support anyone else is a waste because a candidate polling in third or fourth place rarely has any chance of winning a seat. The logic goes that first past the post system should thus lead to two party systems, with a centre-right party competing against a centre-left one. Duverger’s Law when applied to Canada suggests that only one of the Liberals and NDP can play a major role in Canadian politics. If the NDP continues to poll ahead of the Liberals, the Law argues that Liberal voters should move their support to either the Conservatives or the NDP (depending on which of those parties they prefer) in order to keep the party they like least out of government.

More bad news for the Liberal party comes from median voter theory, put forward by Anthony Downs. Downs argues that parties from the centre-right and centre-left should adjust their position to converge on those individuals in the middle of the political spectrum. Voters who are on the extreme of the right and left spectrums will vote for the centre-right and centre-left parties no matter what because they have no alternatives. Centre-right and centre-left parties can thus change their positions without worrying about losing these voters (though there is some threat these voters will stay home, which keeps parties from moving completely to the centre). The voters who are likely to switch parties are the ones in the centre political spectrum, and so centre-right and centre-left parties have strong incentives to move to centre in order to win over centrist voters. As the centre-right and centre-left parties converge they are able to peel support away from a centre party, making it weaker and weaker. The centre party cannot adjust its position to try to fight an encroaching centre-right or centre-left party. Any attempt by the centre party to move in one direction to hold on to its voters means it loses support on the other side of the political spectrum even faster than it would if it did not move at all. In Canada, as the Conservatives and NDP move to the centre the Liberals are increasingly squeezed. Centre-right Liberals are moving over to the Conservatives while centre-left Liberals move to the NDP. If Downs is right, eventually the Conservatives and NDP should be so close that they end up competing over the most centrist Canadian voters, leaving no room for the Liberal party.

The problem with Duverger’s Law and Downs’ median voter theory as applied to Canadian politics (as well as to several other countries) is that they ignore regional and riding politics. The two largest parties in the country are not always the same as the two largest parties in a particular region or riding. When this is the case, strategic voters do not have an incentive to choose between one of the two strongest parties at the national level, they have incentives to choose between the two strongest parties in their riding. This has allowed third and fourth parties to survive in Canadian politics so long as they have significant bases of regional support. Indeed the Bloc Quebecois, the Progressive Party, and the CCF/NDP have been able to win substantial numbers of seats as third and fourth parties because even when they were weak nationally, they had ridings in which they were quite strong. Finishing third nationally will not be enough to force the Liberals out of Canadian politics so long as there are a significant number of ridings in which the Liberals are competitive.

The first past the post electoral system also affects the incentives that parties have to move towards the median voter. In such a system parties are not competing over a single national median voter, but rather over 338 median voters (one in each riding). If one assumes a multi-party system, and that there are at least some voters too principled to strategic vote, one ends up with centre-right and centre-left parties that do not need 50% of the vote to win a majority of seats and that therefore do not need to win the support of the national median voter.  All a party needs to do is get to the 40th or 45th percentile in over 50% of ridings in country- something that can be done without moving completely to centre of the political spectrum. Because these parties also need to satisfy their less moderate supporters (and because parties are democratic institutions where less moderate members can influence party policy), centre-right parties and centre-left parties are likely to leave some room in the middle of the political spectrum that a centre party, like the Liberals, can occupy.

Small levels of support in the centre of the political spectrum may well translate into large amounts of support in particular ridings. It is very difficult to appeal to both small “c” conservative voters in rural Alberta and centrist voters in urban Toronto (or for that matter working class voters and wealthy centrist voters) at the same time. As a result there will always be ridings where either the Conservatives or the NDP is not a viable party. In these ridings the Liberals have a good chance to present themselves as the strongest alternative for voters who do not feel they can support whichever of the centre-right or centre-left parties is strong in the riding. In areas such as urban Toronto, where the Conservatives are uncompetitive, the Liberals can be successful because they can win the support of any voters opposed to the NDP. In downtown Toronto, voters who would otherwise vote Conservative have an incentive to team up with Liberals and elect Liberal candidates in order to keep the NDP from winning the riding. The converse is true for ridings where the NDP is uncompetitive. In ridings such as Vancouver-Quadra, NDP voters have incentives to ally with the Liberals in order to keep the Conservatives from winning. The regional diversity of party support in Canada means that there are a significant number of ridings like this, and suggests that even if the Liberals drop to third that they can still win a significant number of seats.

It is finally important not to write off the importance of third parties when it comes to policy adoption. Strong third parties can play important king-maker roles in parliament, especially in cases where the two strongest parties have similar levels of support. A strong third party can prevent any other parties from winning majority government (as the Progressives did in 1921 and the NDP and Social Credit did for much of the 1960s) and hold the balance of power in minority parliaments. A Liberal party in a minority situation is in a particularly strong position to extract policy concessions from the other parties in parliament. Because it is a centre party, the Liberal party can make a credible claim to align itself with either the Conservative or the NDP. Both parties, thus, have to compete in order to convince the Liberals to back their policy as opposed to their competitors. In contrast, when a party like the NDP is the third party their leverage over policy compromises is limited because the only alliance they can form is with the Liberals. A Liberal party that finishes third may not be able to implement its platform as the lead party in government, but it should still be able to exert a lot of policy influence in any minority parliament.

It remains to be seen whether the Liberal party will be able move back into second place in the 2015 election. The classic political science work by Duverger and Downs suggests that they face an incredibly difficult task as a centre-party trying to compete in a first past the post system. This does not mean that the Liberals cannot play an important role in Canadian politics even if they finish in third place in 2015. They may, however, have to resign themselves to being king-makers as opposed to Canada’s naturally governing party.

* A full discussion of median voter theory can be found in Anthony Downs’ An Economic Theory of Democracy (1957). More on Duverger’s Law can be found in Gary W. Cox’s Making Votes Count: Strategic Coordination in the World’s Electoral Systems (1997).


What Does Voting “No” Mean? Rethinking Referendum Questions

The Greater Vancouver Regional District held a referendum this May on a plan to increase provincial sales tax in order to fund a substantial expansion in public transit and transportation infrastructure. Vancouver voters rejected the tax increase, with 61.7% voting against. This has left municipal politicians scrambling to figure out how they are going to fund a transit expansion that is necessary for a growing city. The result raises questions about whether referendums are an appropriate make decisions about tax and transit policy. My colleague David Moscrop wrote this piece in March arguing that by putting the tax increase to a referendum the BC governments shirked its responsibility to make difficult but important policy decisions. If policy questions are going to be put to referendums though, it may be worth rethinking the way that those questions are framed. The common form of a referendum, a single policy proposition put to a yes or no vote, does not provide an accurate picture of the policy options that governments face nor does it produce a good account of public preferences. Forcing people to make binary decisions on policies prevents them from indicating a preference for alternative policies. Referendums framed as yes or no questions give people the option to say “no” but not to say “let’s try something else.”

As a population we are good at saying “no” to different policy choices. In BC alone referendums have led to the removal of the HST (a referendum to stop a policy initiative the government was implementing) and to the rejection of Single Transferable Vote twice. The binary nature of most referendum questions creates odd coalitions of voters, many of whom prefer different policies. The BC transit referendum, for example, united individuals who opposed raising taxes and the expansion of transit with those who felt that a transit funding increase was necessary, but that a sales tax increase was the wrong way to go about it. One of the objections to the BC government’s plan was that the proposed sales tax increase was regressive. It would be paid disproportionately by low income individuals who spend a greater portion of their income on consumption. Presumably these voters would have been more supportive of an increasing a different, more progressive tax, such as income tax.  Other voters did not want to put more money into transit at all, and would have likely rejected any tax increase regardless of how progressive it was.  The positions of those who oppose tax increases in general and those that want a less regressive tax are very different, yet the referendum treated them as exactly the same. The problem here is that it is unclear what mandate voters have given the BC government and the mayors of the Vancouver region. Should the government cancel the transit expansion, or should they be exploring alternative funding mechanisms? It is clear that Vancouver voters do not want a sales tax increase, but it is not clear what they want in its place.

Perhaps the most concerning effect of the ambiguity of a no vote is that voters may end up with policies that they like less than the sales tax increase that they voted against. Alternative mechanisms for funding transit include raising property taxes or implementing a road pricing (or toll) system. Neither would likely be popular with those that voted no in the referendum because of their opposition to tax hikes. Depending on the degree of property taxes or tolling, its not inconceivable that such a system would cost voters more in taxes than that proposed sales tax increase. Another alternative to the tax increase is simply to decrease funding for public transit and, as a result, service. This is not likely to be a popular option for those that voted against the sales tax because its disproportionate impact on low income individuals. Any decrease to transit service is likely to also disproportionately affect those with low incomes, who are more reliant on public transit. Because the alternative policy preferred by “no” voters is unclear, they can end up with a policy that they prefer less than the one proposed in the referendum.

The alternative to a yes or no referendum would be question with multiple policy options. This would allow voters to distinguish between preferences for a larger set of alternative policies and give governments a clearer picture of what the public wants. It would also protect the public against the danger that a no vote may leave them with a policy that they like less than the one they voted against. It would work like this. Rather than providing voters with a single policy proposal and asking whether they support it, voters would be given four or five different policy options. In the BC transit referendum for example, voters might asked if they support funding a transit increase via a sales tax, via an increase in income taxes, via an increase in property taxes, or if they favoured no increase in transit funding at all. Voters could then rank order their preferences from most preferred to least preferred. When the ballots are counted, the policy with the least support could be eliminated and its votes redistributed to its voters’ second choices until one policy option has more than 50% support. This way, the votes of those that want an increase in transit funding but a different tax from a sales tax can be distinguished from those that want no tax increase at all.

Such a system would give governments a much clearer picture of how the public feels about the policy issue put to the referendum. Rather than being left trying to interpret an ambiguous “no”, provincial and municipal politicians would have a clear indication of the degree to which Vancouverites simply oppose tax increases to fund transit as compared to the degree to which they support the transit funding increase but just prefer it be funded by different means. A clearer outcome would mean clearer policy options for governments and a greater level of confidence in voters that the policy choices they seek in a referendum are the ones that will be put into place.

The Vancouver transit referendum provides an example of how the framing of referendum questions as a binary choice between the acceptance or rejection of a single policy proposal is problematic. Referendum question should be developed in a manner that allows the public multiple options, and gives voters an opportunity to express nuanced views on complex policy decisions. If referendums are going to be used by governments to test public policy, voters ought to at least be given the choice of multiple policy options.