Relentless dynamism - the 2017 election

After a tumultuous campaign, one of the most dramatic in living memory, the election results rolled in on Saturday night, and I thought a few reflections would be useful from the initial voting data. Special votes, some 384,000 of them (15% of overall voting), are currently being counted, and the final result won’t be known for a few weeks. But a few things are worth noting…

Turnout increased... and fell

The number of people voting increased by 110,000 from 2014, up from 2.45 million to 2.56 million. This means that turnout has increased for two elections in a row. Furthermore, the growth in the number of votes cast, at 4.8% between 2014 and 2017 is higher than the growth in the electoral population (3.6%). While this does suggest a more engaged voting population, it wasn’t enough to deliver Labour the breakthrough they were looking for.

And, as Andrew Geddis points out in this piece on coalition prospects, in 2017 only 91.1% of those eligible enrolled to vote, compared to over 93%. He estimates that the turnout of those eligible to vote actually declined from 72.1% to 71.7%.

At least two of the main polls got it right at the end of the campaign

There was a lot of talk in the campaign about the reliability of the polls, fuelled by the contrasting findings coming out between different polling companies. Many queried if landlines were a suitable way to poll people.

In the penultimate week of the campaign, the Newshub-Reid poll had National substantially ahead. At the time, I wrote that it looked like it was an outlier, and the perceived poll volatility led some to go so far as to suggest poll failure. Yet in the final week, the final Colmar-Brunton poll came into line with the Newshub poll, with almost identical results. The final Roy Morgan poll was less accurate, and which in likelihood reflected that it was completed 10 days earlier, and over a longer timeframe.

The table below shows the eve of election poll results published before the election, and the election night initial results. The green cells show which were closest for each party. The closeness of Newshub and TVNZ polls to the election night results are striking. Across all parties, Newshub varied by only 0.8 percentage points on average, and TVNZ by 1 percentage point.

Comparing the main publicly-available eve of election polls with election night results

Comparing the main publicly-available eve of election polls with election night results

Momentum shifted

The shift in the polls in the final two weeks of the campaign was equally striking. National’s scaremongering campaign clearly had a substantial effect on voting intentions, as shown in the next two graphs, updating earlier posts (this time using a non-linear trendline, one of my own take-homes for future blogging). For National, their vote returned to the mid-40s by the end of the campaign.

nat trend 170925.png

Labour’s polling appeared to have peaked in the first week of September, then fell away afterwards. The overall growth in the Labour vote will be pleasing for them (increasing from 24% in July to 36% on election night, and a huge gain from the 2014 election), but the decline in the final two weeks will still be deeply felt. 

lab trend 170925.png

Prospects for the final count

The election night result had National with a projected 58 seats, Labour with 45, New Zealand First with 9, the Greens with 7 and Act with 1. With 61 needed for a majority in the House, New Zealand First have the choice of aligning with National for a comfortable majority (67 seats), or Labour and the Greens for a bare one-seat majority (61 seats).

Over the next two weeks, special votes will be counted; this time around, there are more special votes than ever before, totalling some 384,000. In past elections, special votes tend to favour left-aligned parties. Two excellent pieces of analysis by Graeme Edgeler on Public Address, and Michael Appleton on Twitter, look at the possible impact on the final vote tally if the pattern of special voting in 2014 is repeated.

Their analyses suggest that the final result could increase the number of seats held by Labour and the Greens by two; and a corresponding fall for National. Time will tell.

If this is borne out, it would give Winston two quite feasible coalition options on paper – 65 with National and 63 with Labour and the Greens, both with some level of stability to pass legislation and budgets, and hold a coalition together. At this stage, it is anyone’s guess which way he will go. Because at that point, the coalition becomes less about numbers and more about policy, spending and portfolio concessions – and weighing up if the overall vote is a call for stability or a call for change.

With the largest number of seats, National are claiming victory. However, despite the triumphalist tone of the party on election night, there’s no constitutional obligation in New Zealand to align with the largest party; the Governor-General simply needs to be satisfied that the coalition that has formed has sufficient members to command a majority in the House.

The next two to three weeks will be full of speculation and intrigue, until the next house of cards is eventually built.