Similarities and Differences: What A Comparison of Doug Ford and Donald Trump can Tell Us About Populism in Canada and the United States

Last week produced the election outcome that most expected and many progressives feared.  Doug Ford’s Progressive Conservatives won a large majority in Ontario, taking almost 41% of the vote and 76 out of 124 seats.  There is strong temptation for many to compare Doug Ford to Donald Trump.  Ford is a populist politician, claiming to represent “the people” against a supposed class of political elites.  Such a comparison is valuable in that it can highlight some of the ways that right-wing populists take power.  Like Trump, Ford benefited from running against an unpopular incumbent and for an established party that is usually seen as the alternative to that incumbent.  Unlike Trump, Ford did not run an anti-immigrant campaign.  Indeed, success in the diverse ridings in suburban Toronto were key to his victory.  This suggests that populism can succeed in Canada under circumstances very similar to the conditions its succeeds under in the United States, but not with the same anti-immigrant rhetoric that American right-wing populism often has.

It is notable that both Trump and Ford won elections that a generic opposition party candidate was either favoured to win or very close to being favoured to win.  In the United States, election forecasting models that focus largely on economic conditions and incumbency (and which ignore the characteristics of the candidates running for office) showed a very close election in which a Trump win was reasonably likely.  These models ended up being very close to the actual result.  In Ontario, Ford was running against a highly unpopular Liberal government that had been in power since 2003.  A 15 year old government with a Premier that has an approval rating going into the election of 19% is not one that is likely to win re-election.

Both Trump and Ford won elections that any opposition party should have.  It is very rare to see either the Democrats or the Republicans win three consecutive Presidential terms, as it is to see parties govern provinces for over 15 years.  The take-away here should not necessarily be that Trump or Ford benefited from a populist surge.  Rather, they won elections that most opposition leaders and parties would have.  Progressives should not take too much comfort in this.  One would hope that choosing leaders that ignore democratic norms, like Trump and Ford, would cost opposition parties elections they should otherwise be able to win.  The Ontario election suggests that this is not necessarily the case.

It is important that both Trump and Ford ran for mainstream parties.  This is one of the things that sets the two apart from European populists.  Most European populists run for explicitly populist parties instead of co-opting mainstream ones.  This is the case for Marine Le Pen and the Front National in France, Geert Wilders and the Party for Freedom in the Netherlands, and Nigel Farage and the UK Independence Party in Britain.  As a result, they have to develop their own election organizations and establish their own groups of party loyalists.  Trump and Ford took over existing parties with deep roots in their respective jurisdictions.  Neither had to build their own organization, and both benefited from voters’ loyalty to established parties, be it to the Republicans in the United States or to the Progressive Conservatives in Ontario.

There are two lessons that come from this.  The first is that keeping populist parties out of politics does not necessarily keep populist politicians out of power.  Indeed, institutions that force populist politicians into mainstream parties may make them more powerful.  This is something to consider when debating institutional reforms such as changes to electoral systems.  Designing institutions that make it harder for right win populists parties to emerge may just drive right-wing populists into mainstream parties.  Once in those parties, being able to take advantage of the mainstream parties’ electoral organization and partisan loyalty may make it easier, not harder, for right wing populists to win.

The second lesson is that progressives should not count on partisans to defect in order to defeat a candidate who violates key democratic norms.  There were certainly large numbers of Republicans and Ontario Progressive Conservatives that campaigned against Trump and Ford when they were running in primaries or leadership contests.  Once Trump and Ford had won the leadership of their respective parties, however, supporters of their parties fell in line.  Partisan loyalty ended up being stronger than any commitment to democratic norms.

All of this being said, it is important to remember that Doug Ford is not Donald Trump and that the demographics of Ontario are different from the demographics of the United States.  Unlike Trump, Ford did not make anti-immigrant rhetoric a central part of his campaign.  Rather, he focused on opposing carbon pricing, reducing hydro-electricity rates, and cutting taxes.  Ford did make a rather odd comment about Ontario “taking care of [its] own” in response to a question about encouraging immigrants to move to Northern Ontario.  In responding to criticism though, he emphasized his ties to immigrants and highlighted the importance of recognizing immigrants’ education credentials.  This is far cry from the anti-immigrant rhetoric of the Trump campaign or of many European far-right populists.

Ford also did well in ridings with large shares of immigrants.  The first graph below shows little relationship between the size of the immigrant population in a riding and Progressive Conservative vote share.  The second and third show the average PC vote in ridings with an immigrant population of at least 30% and at least 40% respectively.  Ford did only slightly worse in ridings with an immigrant population of at least 30% than he did in Ontario as a whole.  He actually did slightly better in ridings with immigrant populations above 40% than he did in the rest of the province.

Immigrant Population and PC Vote Share

Each x denotes a riding.  The line shows the trend and the 95% confidence level.

Support in Ridings with Immigration Populations Over 30%

Support in Ridings with Immigration Populations Over 40%

The lack of anti-immigrant rhetoric in Ford’s campaign is a reflection of Canada’s demographics.  Canada has a very large immigrant population (1/5 of Canadians were born outside the country).  Immigrants also tend to be concentrated in swing ridings that determine who wins elections.  This is particularly the case in Ontario with respect to the suburbs around Toronto (often referred to as the GTA).  Running on an anti-immigrant platform in such an electoral environment is bound to lead to failure.  As such, Ford had to modify his populist appeals in a way that appealed to the Ontario’s diverse electorate.

There is value in comparing Trump and Ford’s electoral success.  There are similarities between the two.  Both took advantage of circumstances that were favourable to opposition parties, and both successfully co-opted mainstream parties with strong election organizations and loyal followers.  At the same time, Trump and Ford are not the same kind of populists.  Ford’s efforts to, not only avoid anti-immigrant rhetoric, but to win over immigrants highlights the way that Canadian right-wing populism had to be very different from American right-wing populism.  The anti-immigrant rhetoric of the Trump campaign cannot win in a country where immigrants make up a large portion of the electorate and are concentrated in the key swing districts parties need in order to be successful.


Misreading Polls: Reporting on Ontario Election Polls is Creating More Confusion than Need Be

The Ontario provincial election has turned into quite the horse-race.  A decline in support for Doug Ford’s Progressive Conservatives combined with sustained growth for Andrea Horwath’s NDP has made what was projected to be a PC landslide into a very close election.  With this closeness has come a flurry of news stories claiming (depending on the day and the news outlet) that the NDP has inched into the lead or that the PCs have retaken it.  These stories probably overstate the volatility in the race.  For a week now the race has been tight.  Using polls to suggest anything beyond that, particularly that any party has jumped to a 1 or 2 point lead, overstates the precision one can expect polls to have.  Those watching the Ontario election should be very wary of reporting that fails to highlight margins or error in polls and of reporting that focuses on single polls.

Margins of Error

There is a tendency when reporting on polls to focus on the single number estimate the poll gives for each party.  It is natural to focus on this because it makes for the easiest comparison between parties.  The problem with this approach is that the single number estimate is probably wrong.  Polls are random samples of a population, and the probability that a pollster has managed to come across a random sample that mimics the real population within a percentage is pretty low.  Rather, a pollster using good methods is likely to get a sample that is pretty good but not perfect.  This is why polls have margins of error.  They are an acknowledgement that one can expect a small but potentially important difference between the poll’s estimates and actual public opinion.  A poll that has a margin of error in of plus or minus 3 is essentially saying that it is reasonably likely that its estimate is off by 3 percentage points in either direction.

News articles reporting polls should highlight these margins of error more than they usually do.  Rather than emphasizing the estimated support for a party, reports should make the range the poll gives the party clear.  To do otherwise misleads readers about the level precision and confidence when can have in a poll.  For example, a report that the NDP are polling at 36% suggests that to a reader that the pollster thinks the NDP has 36% of support.  However, if this, imagined, poll has a margin of error of 3, what the pollster has actually found is that the NDP has somewhere between 33% and 39% support.  The poll simply is not precise enough to say more than that with much confidence.

Despite this uncertainty, it is reasonably common to see polls reported on as if their estimates are exact.  This Global news article reports that the NDP are 3 points up on the PCs with no reference to a margin of error.  It goes further and makes seat projections without reporting a margin of error for them (even if one’s polling data is exact, one should expect a second margin of error to exist for any attempt to model seat outcomes from polling data).  This earlier article reports the PCs taking a lead over the NDP even though the 3 point difference between the two parties is less than the 3.2 point margin of error noted at the bottom of the page.  This article should be reporting that it is unclear which party is in the lead.

This kind of reporting would make the most recent week of the Ontario election less interesting.  Instead of getting headlines about how the PCs and the NDP have been taking and re-taking the lead, there would just be a number of reports about how it is impossible to know which party is in the lead.  Reporting that it is uncertain which party is in the lead, however, would be more accurate, and likely better prepare voters for the range of possible outcomes on election night.  It is entirely possible that rather than going back and forth over the last week, the lead changes that polls are reporting are a result of polls’ slightly different samples and methods.  It is also possible that there actually have been lead changes.  Polls are not precise enough to say that either is true.

The Danger of Reporting on One Poll

Even when one takes into account margins of error, individual polls are often wrong.  Sometimes a poll ends up with a weird sample, or a pollster has an odd way of modeling turnout or weighting responses in order to get a sample that reflects the population.  This leads to outlier polls that may be very different from what is actually the case.  If there is only one poll on a particular race, there is always danger that the poll is significantly off either because of randomness that sometimes occurs when sampling is done, or because there is something odd about the pollster’s methodology.

Fortunately, in most elections (particularly federal and provincial ones), there are many different public polls done.  One can compare the results of these polls to arrive at a general trend.  This reduces the chance that an outlier poll will mislead those that read it into thinking support for any one party is different than it actually is.

When there are multiple polls, news organizations should report on polling averages, not on individual polls.  When they do not, they are ignoring information that could provide a clearer picture of what the election actually looks like.  Macleans falls into this trap in this article when it reported on May 29th that the NDP was at 43% (note that this article also leaves margins of error to a note at the very bottom of the page- it is entirely possible their poll had the NDP at 40% or at 46%).  CBC’s poll tracker (which takes a weighted average of polls) for the 29th had the NDP at just 36% (with 43% looking just outside CBC’s margin of error).  Because the CBC average includes a variety of polls, its estimates are based on more information than the Macleans poll (which itself is included in the CBC average) and are probably more acurate.  As much as possible, news reporting should focus on these averages as opposed to individual polls.

Missing the extent to which there is uncertainty in polling can often lead to mistakes when it comes to predicting elections.  In 2016, for example, many (including myself) were surprised by Donald Trump’s victory over Hilary Clinton even though polling average sites such as FiveThirtyEight had a Trump victory well within the margin of error (at 29% the site estimated that a Trump victory was more likely than a coin flip coming up heads twice in a row).  Paying careful attention to margins of error and polling averages may decrease one’s certainty about an election before it happens, but it will also decrease the likelihood that a result is surprising.


Votes and Homes: Why Electoral Reform Matters to Housing Policy in British Columbia

This summer will be an interesting one in BC politics, especially for Attorney General David Eby.  His department is overseeing the development of the rules and question for BC’s upcoming referendum on electoral reform.  Both will play a role in whether the NDP and Green’s efforts to introduce proportional representation to BC will be successful.  At the same time, as the MLA for Vancouver Point Grey, Eby has become the target of protests over the NDP’s new levy on properties worth more than $3 million.  While these issues may seem separate, they are connected.  BC’s current first past the post electoral system makes it essential for the NDP and Liberals to compete over swing districts, including well-off ones like Vancouver Point Grey.  A move to a more proportional electoral system will reduce the importance of these districts, and as a result, the political cost of some of the housing and affordability policies the NDP are pursuing.

Electoral systems shape different voters’ influence over elections.  Under first past the post, where voters live will determine their influence over the election result.  Voters in ridings that one party dominates have limited influence over elections.  No matter who they vote for, it is very likely the dominant party will win the riding.  This is not the case for voters in competitive, or swing, ridings.  In these ridings, the increased competitiveness means a voters’ decision to switch their support from one party to another is more likely to affect which party wins the riding.  In proportional systems, the total number of seats in the legislature a party depends on the party’s vote share across the province, and so where voters live does not change their influence over election results.

The way that electoral systems shape the influence of different voters will affect the political consequences of the NDP’s housing policies.  The party made gains in the last few elections by winning seats in the west and south of Vancouver and in the suburbs.  These ridings include Vancouver Point Grey and Vancouver Fairview (though this riding also has a significant number of low income voters).  The NDP’s future success may depend on the party making gains in similar ridings such as Vancouver False Creek.  Several swing ridings will have significant numbers of home owners likely to be affected by increases in property taxes and other policies designed to bring down housing prices.  Meanwhile, lower housing prices and other programs that make BC’s cities more affordable will disproportionately benefit voters in low income ridings, such as those on the east side of Vancouver (one might think of Vancouver Mount Pleasant and Vancouver Kingsway here).  These are ridings that the NDP already wins by large margins.  Essentially, policies that impose costs on well-off homeowners are likely to hurt the NDP in the swing ridings they need to win power, while winning them votes in ridings they already win by large margins.  These may be good policies, but they are risky politics.

Electoral reform offers the NDP a way out of this problem by reducing the importance of votes in swing ridings relative to those in low income NDP strongholds.  Under such a proportional system, NDP losses in places with large numbers of wealthy home owners, such as Point Grey, could be offset by gains in low income areas of the province such as Mount Pleasant.  This would be the case even if those gains came in parts of the province that they already win by large margins.

This challenge is not unique to British Columbia.  Low income voters in many countries tend to be geographically concentrated in working class urban ridings, while middle and high-income voters tend to be more spread out.  Under first past the post electoral systems, this produces a dynamic where key swing ridings tend to be more well-off than the average and median individuals in a particular country or province*.  As left-wing electoral coalitions shift to include alliances of low-income and well-off cosmopolitan urban voters the extent to which left parties are reliant on urban upper middle-class swing ridings is likely to only increase.  In turn, this will increase the political costs faced by left parties that try to put in place policies that benefit low income voters at the expense of high income ones.

It is unlikely that the NDP are pursuing electoral reform specifically because of the political impacts of their housing and affordability policies.  There is too much uncertainty around the prospects of electoral reform, and the policies are too far apart on the government agenda for this to be the case.  At the same time, these policies highlight the ways in which electoral reform can matter to politics in ways that go beyond determining the numbers of seats each party has in the legislature.  They shape the incentives parties have to respond to different voters and the political risks parties take by implementing different policies.  As a result, it is important to think about the way that electoral reform has implications for policies that impact housing and affordability.

* For more on this it is worth looking at Rodden, Jonathan (2010). “The Geographic Distribution of Political Preferences.” Annual Review of Political Science. 13:321-340.


Did the BC Liberals Get the Right Leader? Comparing the BC Liberal Candidates’ Success in Ridings to Liberal Margins

The race to replace Christy Clark as BC Liberal leader concluded this month with Andrew Wilkinson defeating Dianne Watts on the final ballot.  Both Wilkinson and Watts have connections to the Lower Mainland, where a large number of swing ridings from the last election are located.  Wilkinson is the MLA for Vancouver Quilchena while Watts was the MP for South Surrey White Rock Cloverdale prior to being mayor of Surrey.  This suggests that both have an ability to build Liberal support in one of the more competitive areas of the province.  It is worth examining the relationship between these candidates, as well as the others, with the success of the Liberals in different ridings in the past two elections.  While success in swing ridings in a Liberal leadership race will not necessarily translate into success in a general election, it demonstrates that the leader has support amongst the partisans that will play a vital role in winning the ridings that the Liberals will need in the next election.  The three strongest leadership candidates, Watts, Wilikinson, and Michael Lee all did quite well in the ridings that were competitive in the last election, suggesting that all three will have an important role to play if the Liberals are to be successful in the next election.

Like the federal Conservatives, the BC Liberals use a riding based points system to determine their leader.  Each riding in the province is worth 100 points and each candidate wins 1 point for every 1% of the vote they win in each riding.  The party also uses a preferential ballot.  To win a candidate must have at least 50% of all available points.  If after one round of counting no candidate has a majority of points, the candidate with the fewest points is eliminated and their votes are moved to that candidates’ voters’ second choices.  This system allows me to compare each candidate’s level of support in a riding to the BC Liberal’s margin of victory of defeat in the past two elections (with winning margins counted as positive margins and losing margins counted as negative ones).  Redistricting between the 2017 and 2013 election makes this comparison more difficult for the 2013 election, but fortunately Pundit’s Guide BC has estimated 2013 vote shares for the 2017 riding boundaries.

On the first ballot Michael Lee and Dianne Watts had a significant advantage in close ridings.  The two graphs below show the average support for each candidate in ridings that the Liberals either won by 10% or less or lost by 10% or less.  The first graph looks at ridings which were close in 2017.  The second looks at ridings in which the average margin between the 2017 and 2013 elections was close.  Lee and Watts are more or less tied in ridings that were close in 2017 while Watts is the stronger candidate by two percentage points in the ridings that had a 2017/2013 average margin that was close.  This suggests that it will be important for the Liberals to keep both active in the party in the run up to the next election.  Both were able to mobilize substantial amounts of support in ridings the party will need to win to be competitive in the future.

Support in Ridings Close in 2017

Support in Close Ridings (2017, 2013 avg)

Though the first ballot demonstrates that Lee and Watts did well in Liberal swing ridings, the final ballot shows that Wilkinson also has support in these ridings.  On the final ballot Wilkinson averaged a 1 percentage point lead in ridings that were close in 2017 and almost a 2 percentage point lead in ridings that had a close average margin between 2017 and 2013.  While Wilkinson certainly needs to reach out to Lee and Watts supporters in order to win in competitive ridings, he is not completely dependent upon them.

A comparison of Wilkinson’s and Watts’ support across all ridings shows divergent trends.  The graph below shows Wilkinson’s support was higher in ridings in which the Liberals did better in 2017 compared to ridings in which they did worse.  By contrast, Watts did better in ridings that the Liberals did in worse in than she did in ridings where the Liberals did better.  However, Wilkinson tended to do better than Watts regardless of how well the Liberals did in a riding.  There is a substantial gap between Wilkinson’s and Watts’ vote share in the ridings clustered around the 0 Liberal margin in the graph.  This further demonstrates that Wilkinson has substantial support in competitive ridings.  Watts’ vote shares are also substantial, and so her supporters cannot be ignored, but Wilkinson will not be reliant on them to win close ridings either.  Both sides will have to work together in order to be competitive.

Last Ballot Spport Compared to Liberal Success

Of the three strongest candidates not to make it to the last ballot (I omitted Sam Sullivan because of his low vote share across all ridings), Michael Lee stands out as the strongest in close ridings.  Lee’s support goes up as the Liberal margin of victory decreases, and there is a bit of a gap between Lee and the other two candidates in the ridings that the Liberals either won by a narrow margin or lost.  Both Mike De Jong and Todd Stone did better in ridings where the Liberals were stronger than they did in ridings where the party was weaker.  Stone’s average support only passes Lee’s in ridings where the Liberal won by over 30 percentage points, while De Jong is stronger than the other two candidates in only a handful of ridings.  This highlights the particular need for Wilkinson to reach out to Lee’s supporters.  They make up substantial shares of Liberal supporters in close ridings, while Stone’s and De Jong’s supporters tend to be concentrated in ridings that Liberals already win by substantial margins.

First Ballot Support Compared to Liberal Margin

The BC Liberals were lucky in that all three of their top candidates posted good showing in the ridings that they were close in in 2017 and 2013.  As such, a victory by any of the three candidates would have put them in decent position to contest the next election.  Assuming that the election is contested under a first past the post electoral system (which may not be the case), the party will need supporters of all three to work together in order to ensure the party is competitive in swing ridings across the province.


It’s Complicated: As British Columbia Embarks on a Debate Over Electoral Reform It Is Important to Pay Careful Attention to the Complexity of the Debate

This fall British Columbia will have a referendum on electoral reform.  The government is currently conducting consultations, and both advocates and opponents are making their voices heard on the matter.  Despite the subject’s complexity, nuanced viewpoints are disappointing rare in debates over electoral reform.  Advocates of proportional representation tend to suggest that it can fix all of democracies problems, from increasing voter turnout to increasing women’s and minorities’ representation, to making government’s more representative of the population.  Opponents suggest that proportional representation will ruin democracy, paving the way for the emergence of extremist parties and the creation of legislatures with so many parties that functional government becomes impossible.  Much of this debate misses the complexity that is involved with electoral reform.  No electoral system is perfect, all involve making trade-offs, and the extent to which an electoral system accomplishes any particular goals depends on the details regarding the way the system is designed.  British Columbians should pay careful attention to these trade-offs and details.

An essential thing to consider when debating electoral systems is the trade-offs that need to be made when deciding between systems.  Every electoral system has its costs and benefits.  Advocates of proportional representation often point to disproportionality as one of the central problems with first past the post, and are right to do so.  It is problematic that Justin Trudeau’s Liberals can win a majority government with 39% of the vote under a first past the post system.  The small proportion of the vote needed to win a majority creates further problems as it incentivizes parties to focus their campaigns disproportionately on a narrow groups of swing voters that live in swing ridings.  The fewer the number of voters needed to win a majority, the lower the incentive parties have to run broad-based campaigns that reach across the electorate.  By ensuring that a parties’ vote share equals its share of the national vote, proportional systems give parties incentives to reach out to swing voters across the electorate, not just to those that live in swing districts.

At the same time, first past the post electoral systems ensure voters have local representatives that they can vote out of office and increases the likelihood that majority governments will be elected.  Variations on proportional systems such as mixed member proportional or single transferable vote (STV) can create space for regional representation within proportional systems, but they either dilute regional representation by adding members of parliament (MPs) elected off of party lists or dramatically increase the complexity of the system making it harder for voters to understand.

Because it is rare that a party ever wins more than 50% of the popular vote, proportional systems inevitably reduce the likelihood of the election of a majority government.  There are merits to minority and coalition governments, as they force parties to work together in government.  At the same time, coalitions can be difficult to form if a large number of parties win election to a legislature (as has been the case in both Germany and the Netherlands after recent elections in both countries).  They can also be unstable, as Israeli coalitions often are.  In some cases, minority governments and coalitions work very well, reflecting the views of a broader range of voters better than majority governments do.  In others, they can be unwieldly and unstable.  It is hard to tell which will be the case until one sees how a particular set of parties works under a particular electoral system.

Opponents of proportional representation will often to point the fact that such systems make it easier for extremist, particularly far-right, parties to enter a legislature.  These claims are generally true as proportional systems usually make it easier for small parties of all types to win seats by reducing the number of votes a party needs in order to win their first few seats.  At the same time, one should not over-state the extent to which first past the post guards against such extremism.  The Front National, UK Independence Party (UKIP), and Donald Trump have all demonstrated that far-right parties and candidates can be successful in first past the post or similar systems (France uses a run-off system that is different from first past the post, but is not a proportional system).  Extremist movements that end up forming their own parties in proportional systems often find their ways into mainstream parties in first past the post systems.  The Canadian Conservatives, for example, saw far-right leadership candidates in Kellie Leitch and Steven Blaney.  Significant numbers of Euroskeptic anti-immigrant voters in Britain that may have supported UKIP in a proportional system have found their way into the British Conservative party.  First past the post systems can make it harder for extremist movements to form their own parties and win seats in a legislature, but they cannot erase such views from society nor can they prevent them from having any influence on politics.

Finally, proponents of proportional representation often argue that such systems increase the representation of women and ethnic minorities in legislatures.  This is only partially true.  Whether a proportional system increases the representation of women often depends on the design of the system and the importance different political parties attach to women’s representation.  It is certainly true that countries with proportional systems such as Sweden, Belgium, Denmark, and the Netherlands have some of the highest proportions of women in their national parliaments in the world.  At the same time though, proportional systems in Ireland, Israel, and Slovakia have not kept those countries from having lower levels of women’s representation that non-proportional systems in Canada, France, and the United Kingdom (here is a full ranking of countries with their electoral systems noted).  Ethnic minorities in British Columbia and Canada might have a more difficult time getting elected to parliament if either switches to a proportional system.  First past the post creates districts in places like suburban Vancouver with large numbers of immigrant and ethnic minority voters.  The need to win these seats gives parties strong incentives to be responsive to minorities’ interests and to run minority candidates.  Under a proportional system the number of these ridings would be reduced (in order to allow for the list seats needed to make parties’ seats proportional) or would disappear entirely.

The details regarding how electoral systems work are also important.  There is a tendency amongst both advocates and opponents of proportional representation to divide systems into proportional and non-proportional systems with little reference to how different proportional systems work.  This is problematic because some of the effects of proportional systems will depend greatly on the design of the system.  The extent to which voters will be able to remove MPs they do not like and the level of party discipline will change depending on whether British Columbia adopts an open or closed list system.  Open lists provide voters with the opportunity to choose which MPs will enter parliament for a party, closed lists allow the party to make such a determination.  The decision whether to adopt a party list, mixed member proportional, or single transferable vote system will also shape the incentives MPs have to represent constituents in local ridings.  In list proportional systems, all MPs are elected off of a party list, giving them limited incentives to respond to be concerned with local issues.  In mixed member proportional, around half of MPs come from ridings similar to first past the post ridings, giving some MPs some incentive to be responsive to local concerns.  In single transferable vote, all MPs come from ridings that elect multiple members.  To get elected, MPs must compete, not only with candidates from other parties, but with their own parties as well.  The need to differentiate themselves from other candidates from the same party gives MPs in STV systems particularly strong incentives to be responsive to local concerns and makes it easy for voters to remove MPs they do not like.  Indeed, MPs elected under an STV system may even be more responsive to local concerns than those elected under a first past the post one.

Public debates over electoral systems are challenging.  They are highly complicated, require voters to make careful trade-offs, and require that voters understand highly technical details regarding how different electoral systems work.  At the same time, they are essential.  It is problematic to allow politicians to choose the rules that they compete under with no public input. One of the challenges that British Columbians will face in the run up to the referendum will be carefully understanding how each system works and the trade-offs involved with each electoral system.  In the interests of having a productive debate advocates of different systems should be cognizant of the importance of trade-offs and the details of the different systems when making the case for different electoral systems.


Right/Left Splits: Gains from the Smaller German Parties Came from the Expected Larger Parties

Last September’s German election produced interesting results.  The election created a Bundestag with 6 parties, more than any election since WWII.  This has made coalition formation difficult as the Christian Democratic Union’s (CDU) Angela Merkel failed to build a coalition with the Free Democrats (FDP) and Greens, and is now trying to form a grand coalition with the Social Democrats (SDP).  Part of the reason that coalition formation in Germany is so difficult after this election is that both the CDU and SPD lost significant shares of the vote to small parties.  The CDU dropped from 41.5% of the vote in 2013 to 32.9% in 2017, while the SPD dropped from 25.7% to 20.5%.  In light of this, it is interesting to look at which smaller parties have benefited from the declines in the two major parties.  When one looks at where the CDU and SPD have lost votes, it appears that each of the smaller parties benefited from the larger parties’ decline, though the AFD did particularly well in districts where the CDU or their Bavarian sister party the Christian Social Union (CSU) lost significant amounts of the vote.

I took advantage of Germany’s mixed member proportional system to examine the way that votes shifted from larger to smaller parties in the last election.  Like Australia, Canada, and the United Kingdom, Germany is divided into districts (known in some countries as ridings or constituencies) for elections.  Unlike in Australia, Canada, and the United Kingdom, German voters get to vote both for a district representative and party that they prefer.  The candidate with the most votes in the district enters the Bundestag as the representative for that district, but additional members are also added to the Bundestag from party lists in order to ensure that each party’s total number of seats is equal to its share of the party of vote.  To examine shifts in vote share I looked at the way that voters’ party vote shifted in different districts, comparing the losses in the two main parties to the gains made by the smaller parties.  I used voters’ party vote and not their vote for district representative, because party votes are less likely to be subject to strategic considerations (district representative votes for weak parties are more likely to be seen as wasted than party votes) and because party votes are less likely to be influenced by the quality of the candidate being run in the district.  In all of my analysis, I look at the change in party vote share between the 2017 and 2013 elections.

The strongest differences with respect to where a minor party gained votes exist for the AFD.  The graph below compares the change in vote for the two largest parties to the change in vote for the AFD in each district.  Districts that saw AFD gains between 10 and 20 percentage points saw much larger declines in CDU/CSU vote shares (in black) than SPD vote shares (in red).  Where the AFD’s gains were weaker (between 5 and 10 points), the CDU/CSU’s and SPD’s losses where much closer to each other.  This suggests that the AFD is pulling more votes from the CDU/CSU than from the SPD.  This fits with the way that the AFD has challenged Merkel for her openness to receiving refugees during the Syrian refugees crisis.  At the same time, the fact that the AFD made gains in districts that saw similar CDU/CSU and SPD losses suggests that it is plausible that the AFD took at least some votes from the SPD.

Change in AFD, CDUCSU, and SPD Vote by District

The party with the next greatest divide in gains from the CDU/CSU and SPD is the far-left Die Linke or Left party.  The graph below shows that where Die Linke made gains, CDU/CSU and SPD losses where fairly similar.  CDU/CSU losses were greater than the SPD’s, but that is likely due the fact that the CDU/CSU losses in general where greater than the SPD’s.  This should not be seen as evidence that the Die Linke is drawing a large number of voters from the CDU/CSU.

Change in Die Linke, CDUCSU, and SPD Vote

Interestingly, the districts that Die Linke lost most in are also the districts that the CDU/CSU suffered their heaviest losses.  It is plausible, though more evidence would be needed to prove this, that the AFD has managed to appeal to disillusioned voters on both the left and the right of the political spectrum.  Voters who voted for Die Linke in 2013 out of disillusionment with the mainstream parties may have seen the AFD as plausible alternative to mainstream parties in 2017.  While Die Linke and the AFD have very different anti-establishment appeals, Die Linke’s is economic and anti-capitalist while the AFD’s is anti-immigrant and anti-European integration, it is plausible that anti-establishment voters would shift between the two parties.  This may have occured if AFD was able to increase the salience of immigration and European integration issues at the expense of the economy.  In such circumstances anti-establishment voters may shift from Die Linke as the economic anti-establishment party to the AFD as the anti-immigrant anti-Europe anti-establishment party.  This is supported by the graph below, which shows that the districts that saw the largest declines in the Die Linke vote also saw the largest gains for the AFD.

Changes in Die Linke and AFD Vote

It is finally notable that the differences between who the two mainstream smaller parties took votes from are less pronounced.  The first graph below shows that the FDP did slightly better in districts where the SPD lost more votes and where the CDU/CSU lost slightly less, but these differences do not appear to be particularly significant.  The second graph below shows that the size of SPD losses appeared to have no correlation at all with the change in the Green vote, though the party did do slightly better in districts where CDU/CSU losses where larger (the size of the Green gains in these districts however is only two of three points).

Change in FDP, CDUCSU, and SPD Vote by District

Change in CDUCSU, SPD, and Green Vote by District

The greatest damage done to the two German mainstream parties in 2017 was done by the AFD.  Their growth disproportionately affected the CDU/CSU, suggesting that they are a greater threat to the CDU/CSU than to the SPD.  At the same time, the SPD also lost votes in the districts where the AFD saw gains.  While the SPD should feel less threatened by the AFD than the CDU/CSU, it is not necessarily the case that they face no threat at all from the party.  Far-right parties around Europe have been able to take votes from anti-immigrant left-leaning voters who perceive immigration and the European integration as a threat.  This should lead the AFD to be able to take at least some votes from the SPD, even if they take a larger number from the CDU/CSU.  The gains that the AFD made at the expense of the CDU/CSU mean that, as Angela Merkel attempts to navigate her through coalition talks, she will have to pay particular attention to the threat the AFD presents to her party.


What We Don’t Know: Some Care Has to be Taken When Anticipating the Consequences of Electoral Reform

Electoral reform is on the political agenda again in British Columbia as the Green supported NDP government has committed to holding a referendum on proportional representation (PR) in 2018. As expected, this has produced contentious debate between advocates and opponents of PR.  The examples used by each side of this debate are predictable.  Advocates of PR point to Germany and sometimes Sweden as examples of how such systems can produce stable, effective, and inclusive governments.  Opponents point to countries such as the Netherlands and Israel as examples of how PR can produce large numbers of parties, unstable governments, and coalitions beholden to fringe parties.  There is some truth to both claims.  Debates over electoral reform require individuals to make guesses as to what will happen in future elections.  This, like much of politics, involves making decisions based on uncertain estimates as to what will happen in the future.

A common argument against proportional representation systems is that they lead to the creation of large numbers of parties, unstable coalitions, and governments that can only hold power if they satisfy the wishes of extremist fringe parties.  The Netherlands and Israel are often pointed to as examples of this.  The last Dutch election saw 13 parties win seats in parliament, the largest of which won just 22% of seats.  It then took 225 days to form a coalition government that included four different parties.  Israel’s last election saw 10 parties enter parliament, the largest of which has just 25% of seats, and features a coalition government with 6 parties.  The coalition includes a nationalist pro-settler party (Jewish Home), a secular nationalist party (Yisrael Beiteinu), two moderate right parties (Likud and Kulanu), and two religious parties (Shas and United Torah Judaism).

It is not clear whether proportional representation in BC or federally would create a party system like the Dutch or Israeli ones.  Proponents of PR point out that a threshold can be used to keep out the weakest of small parties.  A threshold of 4%, however, has not prevented Sweden’s last election from producing a parliament with 8 parties and a weak centre-left minority government coalition that is reliant on support from the centre-right to stay in power.  Indeed, coalition instability in PR systems is not as much a result of the entrance of really small parties as it is the weakness of the largest two parties.  The 5 or 6 parties that win between 5% and 15% of the vote are the parties that create the instability in both Dutch and Israeli coalitions.

The emergence of a large number of parties in PR systems is often driven by the extent to which there are a large number of political cleavages in the country.  This is particularly the case in Israel where divisions over left/right politics, security, ethnicity (both between Arabs and Jews and between Ashkenazi, Mizrahi, and Sephardi Jews), and degree of religious observance create space for a wide-range of parties that can all claim to represent different groups of Israelis.

It is hard to imagine that BC and Canadian federal politics would become as divided as Israel, but there may be enough divisions to substantially increase the number of parties competing in elections.  It is plausible that the BC Liberals breaking into a more centrist and fiscally conservative party and a more socially conservative, rural party.  One can also imagine a rural/urban split threatening the NDP’s cohesiveness.  At the federal level regional as well as ideological divides could threaten the stability of Canada’s parties.  The Conservative party could break into more fiscally conservative and socially conservative wings, similar to what was seen with the split between the Progressive Conservatives and Reform in the 1990s.  This could be exacerbated by divides between Western and Ontario/Quebec Conservatives, particularly over issues related to the accommodation of Quebec.  The NDP may also have problems holding together Quebec supporters that might have different views on multiculturalism than the rest of the country and voters that differ on the extent to which they want the NDP to move to the centre in order to win votes.  In both British Columbia and in federal politics it is not hard to imagine the emergence of a far-right party in a PR system given that almost every European country has seen the emergence of such a party.

On the other hand, advocates of PR often point to Germany as a case where PR has produced quite stable governments.  Germany is not an isolated case.  Indeed, many of the countries that are now used as examples of how PR can create unstable governments and party systems have had stable governments under such systems in the past.  Israel was governed by relatively stable Labour led coalitions from its creation in 1948 to 1977.  Italy, which now has so-called “pizza parliaments” with large numbers of parties, was governed by Christian Democrats from the end of WWII to the early 1980s.

There are good reasons to believe that both BC and Canada could end up like Germany.  Canadian parties have a long history of brokering regional and ideological differences.  They are also likely to have a strategic incentive to continue to do this.  Larger parties are more likely to be able to form government.  Even in a PR system, a group of left or right parties that are too fractured may end up conceding government to parties on the other side of the political spectrum.  It is entirely possible that even if regional break-away parties do form that they could be incorporated into permanent alliances with one of the major parties.  This has happened in Germany where the Christian Democratic Union is in a permanent alliance with the Christian Social Union (a party that only runs in Bavaria).  The two parties do not run candidates against each other and always work together in parliament and in government.  Similar arrangements might develop if the NDP breaks into Quebec and rest of Canada factions or if a Western faction were to break away from the Conservatives.

It is finally worth noting that this uncertainty over the future of Canadian parties and government is not limited to proportional representation systems.  Electoral systems do not have to change for party systems to.  Canada saw its party system fracture as a result of regional tensions in the late 1980s and early 1990s.  It is entirely possible that debates over multiculturalism and religious accommodation could create divides between parties in Quebec and the rest of Canada that fracture the Liberal party or NDP, lead to the emergence of a new nationalist party in Quebec, or lead to the revival of the Bloc Quebecois.  The emergence of the Bloc Quebecois and the Reform party in 1993 demonstrates that the first past the post electoral system does not completely protect a party system from insurgent parties (even if it can often make life more difficult for them).  The ability that far-right Republicans such as Donald Trump to gain Republican nominations demonstrates that first past the post systems are not immune to far-right politics.  As much as one should be uncertain about how BC or Canadian politics might change if either adopted a proportional representation electoral system, there is also some uncertainty that would exist if either decides to keep a first past the post electoral system in place.

Uncertainty is a necessary part of politics.  It is impossible to know for certain what the effects of a change in electoral system will be, nor is it possible to say for certain that a first past the post electoral system will ensure stable governments and keep far-right parties out of politics.  With respect to proportional representation there are plausible cases to be made that the adoption of such a system will lead to a stable party system and stable governments similar to Germany (as PR advocates argue), or that such system will lead to more fractured government and party system similar to the Netherlands or Israel (as opponents of PR argue).