Wednesday, November 30, 2016

Do South Australian voters get a Rau deal from the 'Voter Choice' Bill?

On 16 November this year, South Australian Attorney General John Rau introduced a bill into the House of Assembly (the lower house of South Australia's bicameral Parliament) to change the electoral system used to elect members of the Legislative Council. At present, members of the 22-member Legislative Council are elected eleven at a time for eight year terms; half face election at each election for the House of Assembly, the members of which serve four-year terms.

These elections use the single transferable vote system in one statewide district. However, the system has an important modification. Voters can either vote for one party ticket, which accepts an ordering of all the candidates in the election determined by that party and lodged with the Electoral Commission ahead of the election, or number every candidate in order of their preference. I have written about the severe flaws of this system in the past when used at the federal level, and many of the same criticisms apply to the South Australian system.

What the proposal means

If the bill is passed, South Australian voters will be faced with the same ballot paper (reproduced below) that they have received at past elections. However, the law changes substantially the functions of the above-the-line box.
20140102 SG IMG AboveLine
South Australian Legislative Council sample ballot paper (Source: Government of South Australia 2011)
Under the current system, a vote like the one on the sample paper above would have been considered an adoption of the ranking of all the candidates that was submitted to the Electoral Commission before the election. Group C might have asked that an ATL vote for them next go to the candidates of Group A, then Group D, then B and E.

The new system changes the meaning of this vote. Now it only goes to the candidates of Group C. If all the candidates of Group C are excluded, the vote then exhausts. More importantly, voters may only cast one first preference above the line. So even if the voter voted 1 for Group C and 2 for Group A, the vote would only go to the candidates of Group C, and then exhaust. If a voter wishes to express preferences across party lines, they must vote below the line; the law creates a savings provision so that a first preference for the first candidate within a party group counts as an above-the-line vote for that party.

How would this work in practice? Well, the STV system initially allocates seats to candidates who receive a Droop quota ((votes/seats+1)+1). Once no candidate has a Droop quota, the candidate with the lowest number of votes is excluded, and their preferences. However, given that most South Australians vote above-the-line (96% at the last election), and that all above-the-line votes will immediately exhaust when their party does, this will mean that the exclusions will have a very minor impact, and it is most likely that the candidates which have the highest vote share after the process of allocating seats to candidates with Droop quotas will fill those final seats.

In effect, this means that the electoral system will be very much similar to the 'largest-remainder' method of party-list proportional representation. Each elected candidate within a party will receive exactly a quota of that party's vote before being elected, and the last candidate will receive the rest of that party's vote (the remainder).

How would it work?

It's worth noting that the largest-remainder system is designed to produce a generally proportional outcome, so the Legislative Council will likely continue to have a more proportional composition than the House of Assembly, which is elected in single-member districts using the single transferable vote. This means that the chamber can still fulfil its role as a check upon the powers of the Assembly; it may reject legislation, but cannot remove the government.

Nonetheless, the system will switch emphasis from attracting preferences to entirely attracting first preference votes. It would also cause a small change in the composition of the current Legislative Council, as can be seen below.

Elected in 2010
Elected in 2014
Change from GTV
Nick Xenophon Team
Family First
Dignity for Disability
The Dignity for Disability Party won their seat in 2010 off preferences from other small parties, their first preference vote being only 1.2%. The new system would have given that seat to the Liberal Party. Family First won a seat off 4.3% of the primary vote and preferences from other parties; the new system would make that 0.52 quotas, to 1.55 for the Xenophon Team. It is possible that the new system could increase the number of below-the-line votes and make preferences a factor, hence giving that seat to Family First.

Michael Gallagher's 1992 paper on the subject of different methods of party-list proportional representation discusses effective thresholds for winning seats under these methods. He concludes that the threshold of exclusion; that is to say, the minimum number of votes a party can receive without securing a seat (adding one to this total will guarantee that party a seat) for the Droop quota and largest remainder system is 1/(s+1), where s is the number of seats to be allocated (this is in percentage terms). For South Australian elections, that would be equal to 8.33% of the vote. 

Distinct from this is the threshold of representation; that is to say, the smallest number of votes a party can receive and still receive a seat. For this particular figure, the equation is 2/(p(s+1)), where s is seats, and p is the number of competing parties. At the last election, this would be equal to 0.64% of the vote. A more general threshold equation, 75%/(s+1), gives a figure of 6.25%. 

Wasted votes

One advantage of the current STV system is that nearly all votes do go to an elected candidate. Given that a candidate must receive one quota to be elected, and the quota is (votes/(seats+1)+1), only enough candidates to fill all the seats may receive quotas; hence, the maximum number of votes that do not end up with an elected candidate cannot be more than votes/(seats+1) (8.33%).

The new system, however, would be more likely to disregard a larger number of votes. If we consider a vote that is not part of a quota or a remainder rewarded with a seat 'wasted', for my analysis of the last state election, 16.4% of all votes would be wasted, compared to the 8.3% of the vote left with candidates not elected in the actual election. These figures are relatively consistent for recent elections.
Nonetheless, it is worth noting that, to some extent, the current system lowers this figure to an unrealistic amount if voter preferences are to be legitimately consulted. If group voting tickets are to be done away with, getting this figure of only a quota wasted would require voters to cast a preference for every candidate or party group. This would lead to a high informal rate; when voters were required to preference every candidate for the federal Senate, informal rates larger than 10% were not uncommon. Usage of the federal Senate system, which has optional above-the-line preferences and potentially allows votes to exhaust, would increase this figure.

Labor's hypocrisy on exhausted votes

The South Australian Labor government's willingness to tolerate high exhaustion rates contrasts interestingly with Federal Labor's conduct during the debate over introducing optional preferential above-the-line voting. Labor opposed the changes, which passed nonetheless with support from the Greens and Nick Xenophon, on several grounds. One of them was the claim, made by South Australian Senator Penny Wong, that "It (the reform) will mean the votes of up to three million voters effectively going in the bin" and "disenfranchises more than three million voters—people who, at the last election, chose to vote for someone other than the major parties or the Greens". 

Ignoring, of course, what a load of nonsense these predictions were proven to be, I find it quite remarkable that the party that found it so undemocratic that above-the-line votes could exhaust is now introducing a system that guarantees above-the-line votes will exhaust.

 Why the state Labor government is not simply copying the federal system is somewhat confusing, and introducing preferences would improve on this proposal substantially. Nonetheless, the proposal does give voters certainty that their votes will count for their explicit choices only, and the value of preferences under GTV are very much questionable if preferences do not flow the way a voter wishes.

As always, Kevin Bonham has an excellent take on the matter, if you wish to read further.

Monday, November 21, 2016

Republican unified government and the Constitution

If you were surprised by the results of the United States elections, you are not alone. The unexpected election of Republican candidate Donald Trump to the Presidency, and the somewhat less surprising retention of both houses of Congress by the party, could give the Republican Party great control over the agenda in Washington for at least the next two years.

The current count gives Mr Trump 290 electoral votes to 223 for Democrat Hillary Clinton, with Michigan yet to be declared (Mr Trump has a narrow lead). This margin of victory in the College is comfortable enough that one or two 'faithless electors' (members of the Electoral College who do not vote with the candidate they were affiliated with on the ballot paper) could not deny Mr Trump victory. This is despite Mrs Clinton being likely to receive the most votes nationwide.

In the Senate, the Republican Party held marginal seats in Wisconsin, Indiana, Florida, Pennsylvania and North Carolina. The Democrats picked up Illinois and New Hampshire, but this was not enough to defeat the 54-44-2 (independents caucusing with Democrats) majority the Republicans held going in to the election.

Louisiana's Senate seat is yet to be decided; this state holds an election in which all candidates from all parties participate on Election Day; if no candidate secures a majority, a runoff will take place on December 10. This runoff will be between Republican John Kennedy, who won 25% of the vote in the first round, and Democrat Foster Campbell, who won 17%. In total, Republican candidates won 61% of the vote, so it looks likely that Mr Kennedy will win the seat, and the Republican majority will be 52-46-2.

In the House of Representatives, the Republicans appear to have lost seats, going from 247 to a current estimate of 239. Nonetheless, this is still a fairly comfortable majority. It is unclear how many votes the Republicans will end up winning; their House majority in 2012 of was off 47% of the vote compared to 48.4% for the Democrats; the politicised nature of drawing districts for House seats in the United States, combined with strong Republican control over state governments, means that congressional districts in several states have been drawn to apparently favour the Republican Party.

With such an apparently strong position, the Grand Old Party has defied expectations of doom. And now, some have expressed their concern that the Republican Party's strong hold over state legislatures could allow them to amend the Constitution.

How is the United States Constitution amended?

The United States has been noted for having a rigid, hard-to-amend Constitution. The most common method of amendment is approval by two-thirds of both houses of Congress, followed by ratification by 75% of state legislatures.

In this sense, Mr Chu and those expressing similar concerns are correct. Only thirteen state legislatures are controlled by Democrats. Thirty-eight states are needed for ratification (since 75%*50=37.5, and 37 is less than this), and thirteen legislatures failing to ratify would indeed be the minimum required to block legislation. However, most states have bicameral legislatures, and one house would be sufficient to block ratification of an amendment. In this case, the Democrat position looks more secure. Democrats control one chamber in a further three states, meaning that an amendment on purely partisan lines would be somewhat more difficult than simply overturning the one most marginal Democratic legislature.

However, before the amendment process gets to the point of ratification by the States, approval by Congress is required. This would be the point where the content of an amendment would become somewhat relevant.

What sort of amendment would the Republicans wish to pass?

There are a number of amendments that senior members of the Republican Party, as well as Mr Trump, have either suggested their support for or implied their support for. Nonetheless, it is important to note at this point that unlike in Westminster countries, United States political parties are not nearly as internally disciplined. In most cases, registration and identification as a Republican or Democrat is done on voter registration forms, and there is relatively little control by party organisations over who is nominated. This means that there are substantial ideological differences between members of political parties. Various organisations compile 'ideology scores' that show these differences; as can be seen here.

One briefly popular idea for an amendment was the 'Federal Marriage Amendment', which would have required that all marriages in the United States be between a man and a woman. At the time when this amendment was most popular, several states had introduced same-sex marriage, and socially conservative federal Republicans were opposed to this. In the time since then, however, same-sex marriage has been legalised for the entire United States in the Obergefell v. Hodges case, and same-sex marriage has also earned public approval. The President-Elect, too, has expressed disinterest in the issue, describing it as "settled".

Another proposal, though much more on the fringe, is a constitutional amendment to ban abortion. This proposal has not been substantially discussed, but has occasionally been introduced into Congress as a means to overrule the decision of the Supreme Court in Roe v. Wade in 1974 that the United States Constitution protects abortion rights regardless of federal or state law. Nonetheless, such a proposal would be difficult to pass compared to the easier option of appointing judges to the Supreme Court that would rule in favour of abortion restrictions (approval of a judge requires only a majority in the Senate, compared to the sixty-six votes needed for an amendment). The President-Elect has declared himself in favour of this course of action.

A proposal supported less by the Republican Party than by Mr Trump is the idea that libel laws in the United States should be "opened up". Constitutional amendment could be required to achieve this aim, but other members of the Republican Party do not appear to have expressed views on the subject, implying that it is a pet project of Mr Trump's. This would make it very difficult indeed to amend the Constitution, given the non-existent institutional role the President is given in the amendment process.

Partisan ability to amend

Ignoring the actual content of any amendment, what would it take, however, for Republicans to win enough seats in Congress to theoretically amend the Constitution? In the Senate, Republicans would need to gain fourteen seats; in the House, fifty-one.

It has been noted before that in the Senate, the Democrats face a difficult map. Two United States Senators are elected from each state for six-year terms, with staggered terms meaning that a different group of States vote for Senators every two years. Below is a table showing the different Senators that will be up at the 2018 midterm elections for Congress.

As you can see, more than half of the Democratic caucus will be up for election in 2018. This is because 2012, when these Senators were last up for election, was a successful year for the Democrats, as Senators from the same party as the victorious presidential candidate (that year, Democrat Barack Obama) generally do well. Ten Democrats (assuming Mr Trump's narrow lead in Michigan survives) will be up for election in states Mr Trump won; only one Republican is in a similar position.

Nonetheless, a fourteen-seat gain would be a tough ask for the Republicans. That would involve unseating every single Democrat in a state Mr Trump won, as well as four from states won by Mrs Clinton (most of whom are relatively safe) and holding all of their incumbents. Democrats in red states, like Senators Manchin and Heitkamp, also tend to use the internal ideological flexibility given to them allows them to express views more attuned to the views of their states (both are considered to be on the right-wing of their party, which is reflected by the aforementioned ideology scores); Mrs Clinton had to appeal to more mainstream Democratic voters.

In the House, current counting has the fifty-first most marginal seat (assuming a uniform swing, the most Democratic district the Republicans would need to win) as Massachusetts' 9th district, won by Democrat Bill Keating with 56% of the vote with 34% for his Republican opponent. House members are elected for two-year terms, and will all be up for election in 2018. Now, this way of measuring the likelihood of such an upset has its flaws; it does not take into account potential changes in districts, and assumes a uniform anti-Democratic swing, which given internal ideological diversity would probably not happen. But it nonetheless represents the difficulty inherent for such a swing.

All this would be presumably expected to happen in a midterm election. Midterms are generally bad for the party holding the Presidency, since voters may view it as a way to punish or restrain an existing President without throwing them out of office.

Now, clearly, the Republicans could hold on to their unified government at the 2020 presidential and general election. However, more Republicans in the Senate need to be defended at this election, based on a stronger Republican result in 2014. The best prospect for large gains for the Republicans remains the 2018 midterms.


The 2016 election unexpectedly resulted in unified government for the Republican Party, under a controversial and radical leader. A Democratic weakness in the states that developed under Barack Obama's presidency has led some to conclude that this result can not only allow the Republicans to make their preferred legislation, but to amend the Constitution. Nonetheless, these claims are somewhat exaggerated, even at the state level. The likelihood of an amendment passing would depend on its content, but assuming complete Republican unity on such an amendment, the party would need to gain an unfeasible number of seats in order to make changes.