Previous Next

A Populist Exception?

8

The Unexpected Coalition—Challenging the Norms of Government Formation

Jack Vowles

As established in previous chapters, populism and authoritarianism have strong roots in New Zealand political culture and public opinion. Combined, these two attitudinal dimensions are associated with vote choices for New Zealand First, widely recognised as a populist party. However, populism is also associated with vote choice for Labour and is more strongly aligned with left-wing rather than right-wing opinion. Following the 2017 election, the outcome of government-formation negotiations was momentous and surprising: National, the party with the most votes, was excluded from government. A coalition government was formed but its largest party had failed to gain a vote plurality. This presented a potential challenge to both populist and authoritarian values, raising the question of a legitimacy crisis. Therefore, deeper inquiry is required into the normative foundations of the new government’s claim to take office.

This chapter begins with a discussion of the normative principles used to justify and defend government formation and goes on to test their consistency with voters’ preferences for the party to lead the government, voters’ general evaluations of the parties, leadership preferences, evaluations of leaders, issue and policy positions and how these apparently aligned with the positions of the actual and alternative government coalitions that might have formed. A higher-level norm was also called into question by the 2017 outcome: the justification for coalition rather than single-party government, a key objective of electoral system reform in the 1990s. According to standard theory, both populists and authoritarians should be opposed to the principle and practice of coalition government—authoritarians because they prefer strong leaders and populists because they are claimed to be anti-pluralist. One might also expect overall support for coalition government to fall, particularly among National voters. Satisfaction with democracy may also be affected, which would assist to provide some additional detail regarding the minor decline reported in Chapter 3.

Norms and Legitimacy

In political theory and political science, the concept of legitimacy is both confused and complex (Marquez, 2015). Healthy democracies operate on foundations of congruence between norms and outcomes, elite and mass perceptions of which can be loosely characterised as defining legitimacy or its absence. In electoral politics, there exist winners and losers. When the norms associated with winning and losing are confused or contested, public confidence in the democratic process may waver. Losers must accept that they have lost; if they fail to do so, democracy may be at risk (Anderson et al., 2005). Conflict over norms may open the door to authoritarian populism.

Powell (2000) has identified two normative models of democracy: majoritarian and proportional. Most countries’ institutions and practices fall comfortably within one or the other model. When applied to a parliamentary democracy, the majoritarian model identifies the winner as whoever gains a plurality of votes and a majority of seats in the legislature—the winner is also assumed to be a single party. In a proportional system, a single party is unlikely to win a majority of votes alone and must seek the consent of other parties to govern, either by forming a majority coalition with a subset of those parties or negotiating a minority government of some kind. In a random draw of possible outcomes in a proportional system, with all else being equal, a party that is the plurality winner is still more likely than other contenders to form a government. However, there are many scenarios in which the plurality winner may be excluded from office by a coalition of other parties. In this way, plurality ‘losers’ may become government ‘winners’. Here, the norms underlying majoritarian and proportional systems conflict with one another.

New Zealand’s first election under the mixed member proportional (MMP) system occurred in 1996. Between 1996 and 2014, in seven elections, the party who won a vote plurality was able to form a government. In 2017, this run of plurality winners forming governments ceased. The National Party, in power since 2008, was thrust out of office despite winning by far the largest share of the votes. Instead, the Labour Party formed a government in coalition with the New Zealand First Party, giving that party four of 20 Cabinet positions, and with support from the Green Party on confidence and supply. The Greens also took three ministerial positions outside Cabinet.

This is a situation ripe for conflict between norms. In 2017, many New Zealanders still remembered the old single-member plurality (SMP) system. In that year, a little less than two-thirds of people on the electoral rolls were still old enough to have been eligible to vote under the old system. In fact, many still favoured the SMP system over MMP. Populism has long been identified as a central component of New Zealand’s 20th century political culture, as has authoritarianism; these values have affected people’s electoral system preferences and expectations of government in the recent past (Lamare & Vowles, 1996; Vowles, 2011). Authoritarian attitudes lay behind 1996 preferences to retain the SMP system and 2011 preferences to return to it. Meanwhile, contrary to claims of anti-pluralism in populist thinking, populist attitudes can be identified in the arguments for electoral system change. Rather than aspiring towards consensus-based government, reformers were seeking rule by absolute majorities, rather than narrowly based pluralities, and wished to see governments become more likely to keep their promises and heed public opinion (Katz, 1997; Nagel, 1998, p. 265).

It may be that values are changing, particularly among young New Zealanders, who tend to be less authoritarian in their values than the old (Vowles, 2011, p. 141; see Chapter 3). A majority voted to retain the MMP system at a referendum in 2011, in which younger people with no experience of voting under the old rules were significantly more in favour of the new status quo (Karp, 2014). Despite the continued prevalence of plurality winners forming government until 2017, the norms underlying the MMP electoral system are well understood, particularly among the political elite, most members of which have accepted proportional norms.

On election night in 2017, National Party leader and Prime Minster Bill English did claim the ‘moral authority’ to have the first chance to form a government. Given the six-point vote gap between the two major parties, most observers expected that New Zealand First would form a coalition with National, as it had in 1996 (see e.g. Milne, 2017). Nonetheless, Winston Peters, leader of the New Zealand First Party, announced that he would conduct parallel negotiations with both major parties, as he had also done in 1996. Meanwhile, at the special vote count, released two weeks after the election, National lost two seats to the benefit of the Labour and Green parties, adding to the potential majority for the Labour, New Zealand First and Green parties, if they were able to form a government. It was Peters himself, on 19 October, who made the announcement. Bill English’s acceptance of the outcome was clear—he noted that the result was unusual, given his party’s high vote share (44 per cent), but added that he accepted the results (Radio New Zealand, 2017). He went on to congratulate Labour leader Jacinda Ardern on her success.

However, not all on the right and centre-right were so generous. An incoming National Party member of parliament posted on Facebook that ‘MMP was never intended to deny the party that polled highest by far, the win’—a post that was subsequently deleted (MacDonald, 2017). Former ACT party leader Richard Prebble described the new government as ‘a coalition of losers’ and alleged that there had been a ‘coup’. He went on to say that:

The political scientists can tell us it’s legal but the fact remains—it is undemocratic. For the first time in our history, who governs us is not the result of an election but the decision of one man. Jacinda Ardern is Prime Minister in name only. (Prebble, 2017)

Right-wing shock jock radio host and commentator Mike Hosking declared ‘that’s the madness of MMP’ (2017). This theme was also taken up in Australia (NZ shock, 2017).

Nonetheless, the prevailing tone of media commentary ran against this current of criticism. It was conceded that there had been an attitude shift among voters towards a government of change; therefore, Winston Peters and New Zealand First had made a wise decision. National Party pollster David Farrar even expressed relief that his party had not formed a government with New Zealand First, because such a government had the potential to damage National (Farrar, 2017). Bill English graciously accepted the outcome: ‘we all know the rules, we play by them. This is the result … we certainly accept it’ (Radio New Zealand, 2017). The excitement associated with a change of government and the appointment of a new, relatively young female prime minister added to the mood. Many voices within the intellectual community and the commentariat articulated a normative defence of the result. Perhaps the most accessible, in popular terms, was Eva Allan (on Facebook):

Allow me to explain MMP: There’s one mince and cheese pie left in the shop—it costs $5. Bill has $4.50. Jacinda has $3.70. Winston has 70c. James has 60c and David has 5c. No one has enough money to buy the pie by themselves but Jacinda, Winston and James put their money together and buy the pie. Bill gets no pie because he needed 50c but didn’t have any friends to help him pay for the pie. I hope this helps explain things. (cited in Edwards, 2017)

Voter Preferences and the Government Outcome

Except for a few episodes of anecdotal vox pop coverage in the commercial media, the public response, seemingly positive, cannot be estimated, certainly not in any depth. Admittedly, post-election polls indicated a minor further shift to Labour (see Chapter 9) and a further upward boost in the new year following the announcement that Jacinda Ardern was expecting a baby. On first sight, the New Zealand Election Study (NZES) lacks an instrument to directly assess the normative reception of the change of government. However, because the return of questionnaires took place over several weeks, the data available can indirectly compare how people responded before and after the change of government.

Figure 8.1 displays responses to a particularly pointed question with obvious implications for the legitimacy and acceptability of the outcome: ‘On election day 2017, between National and Labour, which one did you most want to be in government?’ Across the whole period of sampling, from just after the election until the end of February, 48 per cent of those who responded to the question in the weighted sample wished to see a National-led government, compared to 42 per cent who preferred Labour. Upon digging deeper into the data, 44 per cent of those voting for the pivotal New Zealand First Party preferred National, compared to 34 per cent who preferred Labour. These numbers should, of course, be treated with caution. For an unweighted number of 235 New Zealand First voters, the confidence interval is plus or minus 6.4 per cent.

Figure 8.1

Figure 8.1: Preferences for a National- or Labour-led government.

Note: The data were re-weighted on party vote to standardise report of vote across pre- and post-government announcements.

Source: New Zealand Election Study (2017).

Even if the confidence intervals were disregarded, these numbers cannot be taken wholly on face value. Cueing effects should be tested for by breaking the data down more deeply, comparing those responding before and after the announcement of the government outcome. That is, we may investigate whether perceptions of the government outcome affected the responses of voters with weak preferences regarding the two main parties, cueing them towards the winner. This cueing could be operative, not only after the announcement but also before it, exaggerating both early preferences for a National-led government and the later preferences for one led by Labour.

Initial investigation, using the standard weight applied to make the sample representative in terms of demography and voting patterns, indicated a ‘flip’, with a Labour-led government leading after the announcement—an apparently significant cueing effect. However, further inquiry leads us to conclude that it is more likely that the formation of the Labour-led government cued National voters in another way, discouraging those who had not yet responded to the survey questionnaire from doing so. Meanwhile, the non-voters in the sample, who tended to prefer a National-led government, were much more likely to respond after 19 October; those responding later still favoured National but were more likely to choose Labour than those responding earlier. Therefore, the dataset had to be weighted further, to adjust for differential response rates among National voters and non-voters between the two periods, including adjustment for the exaggerated effect of cueing on non-voters following 19 October.

Having applied the new weight, Figure 8.1 shows that, after the government announcement, preference for a National-led government remained ahead, but the gap with Labour was no longer statistically significant. While the specific numbers are only indicative, due to even wider confidence intervals, New Zealand First voters may have flipped. They opted 51 to 26 per cent for National before the announcement of the government on 19 October (a difference outside even a wide confidence interval) and 48 to 33 per cent to Labour afterward (a difference just within confidence intervals). However, non-voters, a much larger population group, also shifted towards Labour after 19 October; even then, they still gave National an edge: 46 to 25 for National against Labour, prior to government formation, and 35 to 33 to National, post government formation. These relatively minor cueing effects among non-voters and New Zealand First voters mostly account for the differences before and after 19 October. However, these conclusions rely on a critical assumption—that respondents were reporting their party vote choices correctly (both before and after 19 October). This assumption is reasonable, given that the main source of error in reported votes is non-voters who claim to have voted; however, these errors are already corrected in our data (see Figure A8.1). Therefore, the government outcome presents a legitimacy issue, particularly from the point of view of at least some constructions of populist attitudes. Admittedly, the gap is a narrow one, and the overall distribution is also influenced by the somewhat greater proportion of pre-announcement respondents, as compared to post-announcement responses in the dataset.

Various other questions in the survey provide more relevant data. When asked which political party they ‘liked the most’ on election day, as with the votes cast, the combination of Labour, Green and New Zealand First narrowly edged out the combined preferences of the parties in the previous government—National, ACT, Māori and United Future—however, by less than 1 percentage point, which is well within the error margin. When asked ‘which party if any best represents your views?’, nearly 40 per cent indicated the parties of the new government, compared to just over 37 per cent choosing the parties of the old: again, this difference is well within sampling error. After being asked to indicate the most important issue in the election, respondents were asked ‘which party would be best at dealing with that issue?’ Across the sample, just over 42 per cent indicated a new government party and just under 35 per cent indicated an old government party. However, this question is also likely to be strongly cued by whether the designated party was expected to be in government and, therefore, have the opportunity to ‘deal’. Before 19 October, with the government formation weight applied, the two sets of parties were level, with the Labour-led parties only marginally ahead, at approximately the 40 per cent mark: after 29 October, the new government’s parties moved decisively ahead at 46 to 30.

Another approach is to calculate the distance between the alternative sets of government parties, the average and the median voter. For this, as shown in Figure 8.2, one relies on the 0–10 left–right scale, where respondents placed themselves and the parties, respectively. From this, we can calculate the perceived left–right position of the government, weighting respondents’ left–right perceptions of the parties by the parties’ vote shares. To include all respondents, missing values and ‘don’t knows’ must be set at the midpoint of 5. Parallel analysis that drops those cases loses many who could not place the smaller parties; however, when the numbers are recalculated on this basis, the gaps stay relatively in proportion. Coding the missing values as 5 pulls all parties towards the centre—the more so the more values that are missing. Figure 8.2 displays these data with the missing values coded as 5. An alternative figure without missing values makes no difference to the findings in Figure 8.2.

Figure 8.2

Figure 8.2: Party, alternative government and respondent right–left positioning and distance from the average and median voters.

Source: New Zealand Election Study (2017).

In this figure, most left is 0 and most right is 10. For the most part, when averaged across respondents, perceptions of party positions accord with expert assessments. The exception is ACT—it scores as less to the right than National. This is because only two-thirds of the sample rated ACT on the scale. Without missing and ‘don’t know’ values set at 5, ACT scores 7.24, only slightly further to the right than National (7.15), also without missing values. The average respondent scores 5.28 (without missing values, only slightly higher at 5.35). The median respondent is 5.

In the government mean and distance rows, left-centre represents the Labour–New Zealand First–Green combination and right represents National, ACT and Māori. Former National support party United Future was not scored on the left–right scale and could not be included, although the party received a microscopic vote in 2017. Right–centre represents a hypothetical National–New Zealand First coalition. The government mean scores put the current left-centre government at just below 4. Right or right-centre differ little, with New Zealand First potentially pulling a right-led government marginally towards the centre. Distances from respondents measure the average difference between respondents’ self-positioning and where they placed the three alternative governments—a more finely grained comparison than that based on the average party positions displayed in the top two sections of the figure. The three alternative governments are almost indistinguishable on this distance estimate. The difference emerges when respondents’ left–right assessments of the three alternative governments are compared with the position of the median voter (at 5). On this basis, the current Labour–New Zealand First–Green grouping is significantly closer to the median. This remains true when missing and ‘don’t know’ values are dropped from the calculations.

On policy and attitudinal grounds, a case can be made that the parties of the new government had a slight edge in terms of public preferences. On grounds of incumbency and perceptions of competence, preferences were more likely to favour the National-led grouping. However, regarding preferred prime minister, less than 2 percentage points gave the edge to preferences for National, ACT and Māori Party leaders against those for Labour, Green and New Zealand First. On ratings of competence and trust, as applied to the two major party leaders, translated into a scale from 0 to 4, the gap is similarly narrow. We have already seen this data broken down by gender (see Chapter 6, Figure 6.2). Across both genders, Figure 8.3 shows that, while English was perceived as significantly more competent, Ardern out-rated him by a somewhat greater margin on trust. Ardern had the advantage of being a new leader, whereas English’s long record as a politician was bound to generate distrust, particularly among his opponents. In Ardern’s case, respondents might have selected ‘don’t know’, again scored as a mid-point in the scale along with missing values, more frequently; however, she was ahead of English, even in the positive responses to the trust question.

Figure 8.3

Figure 8.3: Jacinda Ardern and Bill English—how trust and competence can be used to describe them.

Source: New Zealand Election Study (2017).

On other questions, though, the competence factor weighs in. Regarding the general performance of the previous government, more than two-thirds of respondents thought it had done a very good or fairly good job, with most (54 per cent) selecting the ‘fairly good’ option.

Figure 8.4

Figure 8.4: The state of the economy over the last 12 months prior to the 2014 and 2017 elections.

Source: New Zealand Election Study (2017).

Turning to the economy, Figure 8.4 shows that those expecting things to improve a little were a smaller group in 2017 than 2014, matched by a proportionately larger group in the ‘stay the same’ category in 2014. The outrightly negative groups were equivalent in size in both election years, most likely prompted more by their partisan choices than objective assessment. However, some of the shine of the National-led government’s reputation for ‘good’ economic outcomes was missing when compared to 2014.

On the balance of assessments, one can understand the response of the National Party and its allies to the formation of the Labour, New Zealand First and Green–backed government and give due respect to National’s moderate and gracious acknowledgment of defeat. On election day, by a near margin, a plurality of voters might have preferred, and most expected, that a National-led government would prevail. A similarly small plurality preferred a National Party prime minister. By a significant margin, the electorate rated the National-led government’s performance as positive, and had confidence in the state of economy, albeit with slightly less enthusiasm than in 2014. Equally marginally, it seems that pluralities of voters preferred the issue positions of the new government’s parties to those of its predecessor. Of perhaps greater significance, perceptions of the new government’s position on the left–right dimension put that government significantly closer to the median voter than would have been the case under the parties of the previous government, or even a hypothetical National–New Zealand First coalition. However, this leaves the debate regarding ‘legitimacy’ virtually where it begins, surrounding a very close election and an unexpected outcome that is still questioned by many.

Attitudes towards Coalitions

In a situation where those of the losing side come to question an election result, trust in the political process and in democracy may be damaged. Scepticism regarding the normative basis of coalition governments compared to those of single parties may also increase. Such a situation could encourage authoritarians and populists to further intensify their disdain regarding the values of political compromise and minority positions. If so, how the government outcome affected more fundamental attitudes to New Zealand democracy is a crucial question to address.

One of the key elements of a proportional electoral system is its tendency to encourage coalition rather than single-party governments. Indeed, a preference for coalition rather than single-party government underpinned many New Zealanders’ selection of the MMP system in 1993 (Lamare & Vowles, 1996) and again in 2011 (Karp, 2014). Since 1993, over nine elections up to 2017, the NZES has asked a question with four components, as follows: ‘Generally speaking, do you think that a government formed by one party, or formed by more than one party, is better at doing the following things: providing stability, making tough decisions, keeping promises, and doing what the people want?’ Respondents answer the question for each of its components by indicating ‘one party best’, ‘more than one party best’, ‘both about the same’ or ‘don’t know’.

The data extracted from these questions indicate that most respondents find it easy to distinguish between the advantages and disadvantages of various aspects of performance when comparing multi-party and single-party governments (see Table A8.1). The broad patterns of responses have become relatively consistent, particularly from 2002 onward, as experience has accumulated. Pluralities, and sometimes outright majorities, believe single-party governments to be more stable, although support for each type of government was evenly split in 2014. In 1999, a plurality found single-party governments better at making tough decisions than coalitions. In 2002, there was equal support for both forms of government in terms of being tough. From 2005, pluralities found single-party governments tougher until 2011; however, in both 2014 and 2017, the judgment shifted to coalitions. There is consistent plurality support for coalitions being better than one-party governments at keeping promises and consistent majority support for coalitions being more responsive to public opinion than single-party governments. On balance, and assuming all four aspects of government performance are equally weighted, preferences for coalition government consistently win out over single-party government (Vowles, 2011).

However, there is a complication—popular understandings of coalition are not consistent with the formal definition. Strictly speaking, a coalition government is one in which two or more parties have ministerial positions in Cabinet, the central decision-making body in the executive. All members of Cabinet are bound by collective Cabinet responsibility and are held accountable for all government decisions. In terms of Cabinet composition, coalition governments were formed after the 1996, 1999, 2002, 2005 and 2017 elections. However, following the 2008, 2011 and 2014 elections, the National Party formed single-party Cabinets. Technically speaking, all New Zealand governments since 1998 have also been minority governments who relied for their majorities on confidence and supply agreements with other parties. From 2005 onwards, confidence and supply agreements have been underpinned by support parties usually taking ministerial positions outside Cabinet. Those ministers are responsible for government policy within their own portfolios but are not required to defend or support policy in other areas unless they agree or have committed to do so by agreement. The popular understanding of coalition in New Zealand encompasses this type of government; support parties are considered part of the government, as indeed they are, particularly if their leaders hold significant ministerial portfolios.

The key point is this, New Zealand’s post-MMP experience of coalition government has tended to be shaped by the continued dominance of one large party within a coalition. At times, coalition partners with Cabinet seats have been significant players—from 1996 to 1998, National with New Zealand First, and from 1999 to 2002, Labour with the Alliance. Following 2002 and until 2008, the Labour-led government was in a commanding position. Between 2008 and 2017, a National-led government was in an even stronger position, with three small support partners, not all required for a majority. New Zealanders quite reasonably understand such governments to be coalitions because the core party cannot command a majority on its own and has to negotiate with its partners to achieve its aims, sometimes failing to achieve all that it wants, but usually prevailing. The reappearance of coalition government in its more precise definition in 2017 could therefore shift perceptions, as constraints on the largest party have clearly increased. Indeed, as argued above, the very formation of the Labour–New Zealand First Cabinet provoked challenges to its legitimacy.

To probe the data more deeply, the first step is to deconstruct general attitudes towards coalitions versus single-party government across the populist, authoritarian and left–right attitudinal dimensions in multivariate analysis—the latter to control for the effects of the left-leaning tendencies of most New Zealand populists. For good measure, a control for political efficacy may be added (see Table A8.2). Derived from this model, Figure 8.5 confirms that populists are significantly more likely to be in favour of coalition government in New Zealand than non-populists, even after controlling for left–right position and political efficacy. An alternative model adds age, which is non-significant (whether estimated continuously, or in cohorts). This throws further doubt on any claim that New Zealand populists might be significantly ‘anti-pluralist’—this is despite the retention of an ‘anti-pluralist’ question in the populist scale, indicating opposition to political compromise. Meanwhile, authoritarians are more in favour of single-party government, as expected, and in confirmation of previous findings, with the left being more in favour of coalitions and the right more likely to be against them.

Figure 8.5

Figure 8.5: Associations between populism, authoritarianism and left–right position on preferences for single party or multi-party governments.

Note: The model also controls for political efficacy. Preferences for one-party–multi-party government range between 0 (strongest preferences for single-party) and 1 (strongest preferences for multi-party).

Source: New Zealand Election Study (2017) (see Table A8.2).

Turning to the potential short-term effects of the unexpected coalition, Figure 8.6 shows that support for coalition government drops between 2014 and the immediate post-election period, and again after the formation of the coalition government, while remaining just in positive territory. Meanwhile, Figure 8.7 shows that, measured as a five-point scale between 0 and 1, satisfaction with democracy remained at the same level as in 2014 (before government formation in 2017) and only dropped marginally thereafter. If the unexpected formation of the government had any effects on overall satisfaction with democracy, these were very minor.

Figure 8.6

Figure 8.6: Coalition versus one-party government summary scale (2014–2017).

Source: New Zealand Election Study (2017).

Figure 8.7

Figure 8.7: Satisfaction with democracy (2014–2017).

Source: New Zealand Election Study (2014, 2017).

Figure 8.8

Figure 8.8: One-party versus multi-party government and party votes.

Source: New Zealand Election Study (2017) (see Table A8.4 [Model 1]).

The drop in support for coalition government merits the most attention. One would expect the effect to be found mainly among National party voters—and, indeed, it is. A further regression model included response pre- or post-government formation and three voting categories: those for the three government parties, those for National and those for non-voters and the rest (weighted to correct for response rate bias between the two periods, as explained earlier). With their key effects in the background, the two variables were interacted. Even among government party voters, support for coalition goes down marginally, but remains well within confidence intervals. As Figure 8.8 shows, National party voters drop eight points (from 46 to 38 on the coalition/one-party government scale), just outside confidence intervals.

An age cohort effect was also to be expected. Support for MMP and for coalitions is stronger among those who never voted under the old system and weaker among those who voted under the old SMP system (Karp, 2014). Figure 8.9 displays the results of post-estimation from a regression model with an interaction between government formation and pre- and post-MMP generations, in addition to their key effects. It shows that the pre-MMP cohort drop significantly in their support for coalitions; however, the post-MMP age cohort appears unaffected by the formation of an unexpected government.

Figure 8.9

Figure 8.9: Coalition/one-party government preferences and age cohorts.

Source: New Zealand Election Study (2017) (see Table A8.4 [Model 2]).

In the short term, the intensification of coalition-style politics under the Labour–New Zealand First–Green government could be expected to challenge the positive perceptions of coalition governments that have, until now, been exhibited by the majority of New Zealanders. Indeed, preferences and evaluations of multi-party and single-party government alternatives have waxed and waned over the MMP period. When coalition politics became more conflicted or unstable, opinions shifted, first in 1998–1999, when the first MMP coalition government splintered (Karp & Bowler, 2001), and again in 2005 (Vowles, 2008), when New Zealand First became a support partner for Labour following a very close election and Winston Peters took the position of Minister of Foreign Affairs. A minor age-cohort effect is associated with the short-term shift in sentiment against coalitions among the pre-MPP generation, with no effect at all among those who began voting after the MMP system was established. However, over the longer term, age (however it is measured) demonstrates no association with one-party versus multi-party government preferences after populist, authoritarian and left–right attitudes are taken into account. Even dropping political efficacy from the Figure 8.5 and Table A8.2 analyses does not bring age back into contention. However, doubts remain. Other research provides evidence that, following more extreme conditions of government instability and collapse, as in the aftermath of the events of 1998, young people may react both more strongly and more negatively than older people (Vowles, 2011, p. 141). More robust methods of age–period cohort analysis will be required to investigate these questions more thoroughly (to the extent that this is possible, given the fewer questions available for analysis in earlier data).

Conclusions

The unexpected outcome of the government-formation process in 2017 did not pose a serious challenge to the legitimacy of the coalition government or to the election system that made it possible. On the margins, when they were voting, more people probably preferred the outcome of a National-led to a Labour-led government; however, these margins were thin and flexible—some, at least, adjusted their preferences after the government was announced. The strongest reason to choose National was perceptions of the party’s positive performance in office, whereas the strongest reason to choose Labour or one of its partners was the party’s policy positions. When measured in terms of perceptions of the left–right positions of alternative governments, Labour, New Zealand First and Green were closer to the median voter than the other National-led alternatives.

Satisfaction with democracy hardly shifted—the downward shift was too minor to provoke concern. Support for the idea of coalition governments fell back on the margins but remained significantly stronger than that for single-party government. Meanwhile, populists emerged as the strongest supporters of multi-party government, bringing the association between populism and anti-pluralism into question, at least in the New Zealand context.

In taking stock, as the 2020 election grows nearer, questions of legitimacy remain on the table. A government that is based on three significant parties presents far greater potential for disagreement, division and potential failure than the New Zealand governments formed since 1996. The very composition of the government is a point of vulnerability at which both the Opposition and critical political commentators continue to probe and poke. Evidence of poor performance and, even more seriously, a government collapse, may bring the legitimacy question to the fore. Alternatively, successful completion of the three-party government’s term could cause remaining concerns regarding legitimacy to subside, thus further reducing their relevance in the years to come.

References

Anderson, C. J., Blais, A., Bowler, S., Donovan, T. & Listhaug, O. (2005). Losers’ consent: Elections and democratic legitimacy. Cambridge, United Kingdom: Cambridge University Press.

Edwards, B. (2017). Political roundup: The legitimacy of the Labour-led government. New Zealand Herald. Retrieved from www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11936270

Farrar, D. (2017). It’s Labour. Retrieved from www.kiwiblog.co.nz/2017/10/its_labour.html

Hosking, M. (2017). Mike Hosking: Already a mess—We are all in trouble. New Zealand Herald. Retrieved from www.nzherald.co.nz/nz/news/article.cfm?c_id=1&objectid=11934989

Karp, J. A. (2014). Generations and the referendum on MMP. In J. Vowles (Ed.), The new electoral politics in New Zealand (pp. 187–198). Wellington, New Zealand: Institute for Governance and Policy Studies.

Karp, J. A. & Bowler, S. (2001). Coalition government and satisfaction with democracy: Explaining New Zealand’s reaction to proportional representation. European Journal of Political Research, 40(1), 57–79. doi.org/10.1111/1475-6765.00589

Katz, R. S. (1997). Democracy and elections. New York: Oxford University Press.

Lamare, J. & Vowles, J. (1996). Party interests, public opinion and institutional preferences: Electoral system change in New Zealand. Australian Journal of Political Science, 31(3), 321–346. doi.org/10.1080/10361149651085

MacDonald, L. (2017). MMP attacked online after coalition formed. Newshub. Retrieved from www.newshub.co.nz/home/election/2017/10/mmp-attacked-online-after-coalition-formed.html

Marquez, X. (2015). The irrelevance of legitimacy. Political Studies, 64(1), 19–34. doi.org/10.1111/1467-9248.12202

Milne, J. (2017). Voters cannot, and will not, tolerate Winston abusing his kingmaker position. Sunday-Star Times. Retrieved from www.stuff.co.nz/national/politics/96867393/jonathan-milne-voters-cannot-and-will-not-tolerate-winston-abusing-his-kingmaker-position

Nagel, J. (1998). Social choice in a pluralitarian democracy: The politics of market liberalisation in New Zealand. British Journal of Political Science, 28(2), 223–267. doi.org/10.1017/S0007123498000155

New Zealand Election Study. (2014). New Zealand Election Study [dataset]. Retrieved from www.nzes.org/exec/show/data

New Zealand Election Study. (2017). New Zealand Election Study [dataset]. Retrieved from www.nzes.org/exec/show/data

NZ shock: Losers take power. (2017). The Australian. Retrieved from www.theaustralian.com.au/nation/world/new-zealand-shock-losers-labour-and-nz-first-take-power/news-story/78dfb678806601e8387b2f6c1be3b3ac

Powell, G. B. (2000). Elections as instruments of democracy. New Haven, CT: Yale University Press.

Prebble, R. (2017). Jacinda Ardern will regret this coalition of losers. New Zealand Herald. Retrieved from www.nzherald.co.nz/opinion/news/article.cfm?c_id=466&objectid=11935125

Radio New Zealand. (2017). Ardern’s rise ‘remarkable’—Bill English concedes. Radio New Zealand. Retrieved from www.radionz.co.nz/news/political/341964/ardern-s-rise-remarkable-bill-english-concedes

Vowles, J. (2008). The genie in the bottle: Is New Zealand’s MMP system here to stay? In M. Frances & J. Tully (Eds), In the public interest: Essays in honour of Professor Keith Jackson (pp. 105–125). Christchurch, New Zealand: University of Canterbury Press.

Vowles, J. (2011). Why voters prefer coalitions: Rationality or norms? Political Science, 63, 126–145. doi.org/10.1177/0032318711403917

Appendix

Figure A8.1 reports the vote shares for the four main parties, reported both before and after government formation. Our main concern was the possibility of the misreporting of vote choices among those voting for National and Labour—Labour voters saying they voted National before government formation and National voters saying they voted Labour afterward. On inspection, the figure suggests that the most significant reason for the difference in government outcome preferences lies in the significantly lower number of National Party voters responding to the survey after the announcement. Error in reporting voting choice is less likely an explanation—the difference is too great for this explanation to be credible, except on the margins.

Figure A8.1

Figure A8.1: Reported and validated votes (before and after government formation).

Note: Standard weight applied.

Source: New Zealand Election Study (2017).

While there is a recorded tendency of survey respondents to incorrectly report voting for the winner in post-election surveys, this is most likely to apply to non-voters who report voting. However, such cases are correctly classified as non-voters in our data. There was a much higher proportion of non-voters responding later in the period, but there is no reason to suspect that this had anything to do with the formation of the government—we might expect non-voters less engaged with the election to respond later, if at all. The same propensity for non-voters to respond late was observed in 2014, although the tendency was somewhat stronger in 2017. Non-voter preferences remained somewhat greater for a National-led government throughout, although the margin narrowed to almost zero post government formation. Green and New Zealand First differences were too minor to be significant.

Therefore, it is reasonable to assume that the differences in prior to and post government formation reported behaviour in the initial analysis relate more to the drop in National voting respondents post government formation and the differential response rate of non-voters prior to and post government formation. Misreporting vote choice is likely to be marginal. However, more robust statistical tests for this assumption could be applied.

Table A8.1: Evaluations of coalition versus single-party governments (1993–2017)

Providing stability

One party best

More than one best

Both same

Don’t know

N

1993

45

34

8

13

1,978

1996

46

35

10

10

3,999

1999

57

25

11

8

4,885

2002

56

26

10

8

4,609

2005

61

24

8

7

2,787

2008

48

34

8

10

2,619

2011

47

35

9

10

2,399

2014

39

39

13

8

2,782

2017

46

33

13

9

3,308

Making tough decisions

One party best

More than one best

Both same

Don’t know

N

1993

36

46

8

11

1,959

1996

35

47

9

9

3,961

1999

42

40

11

7

4,851

2002

42

42

10

7

4,594

2005

46

39

8

6

2,767

2008

43

40

8

9

2,606

2011

43

39

9

9

2,374

2014

35

46

13

8

2,882

2017

36

45

11

8

3,298

Keeping promises

One party best

More than one best

Both same

Don’t know

N

1993

17

59

13

11

1,969

1996

20

59

12

10

3,996

1999

30

43

17

9

4,861

2002

29

47

16

9

4,584

2005

38

42

12

8

2,760

2008

35

42

11

10

2,614

2011

31

47

12

10

2,375

2014

26

49

16

9

2,768

2017

33

44

15

9

3,284

Doing what the people want

One party best

More than one best

Both same

Don’t know

N

1993

13

64

12

12

1,973

1996

10

63

15

12

3,992

1999

19

51

19

11

4,850

2002

18

56

17

9

4,598

2005

24

53

14

9

2,763

2008

23

54

12

11

2,608

2011

20

55

13

12

2,370

2014

15

56

18

10

2,768

2017

19

54

16

10

3,301

Source: New Zealand Election Study (2017).

Table A8.2: Populism, authoritarianism and coalition/one-party government preferences

1

2

Authoritarianism

–0.290***

–0.297***

(0.049)

(0.048)

Populism

0.209***

0.211***

(0.052)

(0.051)

Right–left position

–0.245***

–0.250***

(0.049)

(0.051)

Political efficacy

0.087**

0.080**

(0.039)

(0.039)

Age

0.001

(0.000)

Constant

0.650***

0.633***

(0.043)

(0.043)

Observations

3,455.000

3,455.000

R-squared

0.093

0.093

Note: Robust standard errors in parentheses.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Source: New Zealand Election Study (2017).

Table A8.3: Attitudes to single or multi-party government: Vote choice and government formation (Model 1); generations and government formation (Model 2)

1

2

After = 1; before = 0

–0.026

–0.056***

(0.018)

(0.021)

Reference: Government party voter

National voter

–0.198***

(0.018)

Non-voters and other voters

–0.121***

(0.037)

After x national voter

–0.051*

(0.029)

After x non-vote/other

0.044

(0.048)

Post–mixed member proportional (MMP) generation = 1; pre-MMP = 0

–0.002

(0.026)

Before/after * pre/post-MMP generations

0.057

(0.036)

Constant

0.659***

0.562***

(0.011)

(0.014)

Observations

3,455.000

3,352.000

R-squared

0.090

0.005

Note: Robust standard errors in parentheses.

*** p < 0.01, ** p < 0.05, * p < 0.1.

Source: New Zealand Election Study (2017).


Previous Next