Wednesday, 4 July 2018

Farrar Deja vu

Budget day at the House

[White background version here]


In a faint echo of his highly influential Honeymoon Scam, prominent National Party operative David Farrar has made a fresh series of fraudulent Poll Bounce claims on Kiwiblog.

Published in the immediate wake of the latest round of polling, these claims are encapsulated by the title of the post in which they appear: No budget bounce for Labour (28 May) (here)

They are probably best understood as the latest installment in an on-going de-stabilisation campaign to cast the Ardern Government as uniquely fragile, unpopular and illegitimate.

Farrar makes five closely-related but ultimately distinct empirical claims:

His First Contention:
It is remarkable that National remains ahead of Labour considering they are in opposition.
This almost euphoric exclamation was then faithfully echoed, with only minor re-wording (and no attribution), by Jane Clifton in her Listener Politics column (9 June) (here):
National is still out-polling Labour - a remarkable achievement for a first-term Opposition

I'll deal with this contention in greater depth in an upcoming post - one that scrutinizes very similar arguments made recently by Chris Trotter and Heather du Plessis-Allan, albeit couched in slightly different terms  (a post I've been working on in bits and pieces for a few weeks now).

But, for the moment, let me just undercut Farrar and Clifton's expression of sheer amazement by laying out a particularly telling comparison:

  • In the most recent TV1 Colmar Bruntons, National led Labour by 1 point (April 2018: 7 months out from the last Election) and by 2 points (May 2018: 8 months out).
  • Back in 1991, Mike Moore's Opposition Labour Party was already 6 points ahead of Jim Bolger's National Government just 6 months on from National's 1990 landslide victory (which the latter had won by a whopping 13 points). By the late August and September 1991 TV1 Heylen polls (equivalent to July / Aug 2018 in terms of time elapsed since the respective elections) Labour was enjoying a massive 20-22 percentage point advantage over the Bolger Govt. And Moore's Labour continued to easily out-poll the National Administration for almost the entire remainder of the First Term, only slipping behind towards the end.

(And, of course, this is before we even get on to the fact that, by accepting Farrar's FPP-style National vs Labour framing, we are very much buying into an on-going sleight-of-hand (happily embraced by an influential segment of the media) that inherently flatters the Right's sole political vehicle - National - and thus obscures far more than it reveals) (1)

Farrar's Second Claim :
It is remarkable that National remains ahead of Labour considering they (National) ... have transitioned away from the leadership team of the last decade
Farrar repeated this assertion (albeit with slight variation) in a more recent Kiwiblog post: Hooton on National's polling (6 June) (here):
It is astonishing that National continues to poll so high, despite ... a leadership change.

On this specific point, I'll just say the following:

Didn't Farrar - in the immediate wake of Ardern's election as Labour leader in August 2017 - explicitly state (here) that New Major Party Leaders (the most recent example, of course, being National's Simon Bridges) almost always generate a boost for their Party in the Polls ? As opposed to triggering a fall in support ?

In other words, pretty much the polar opposite to the implications of his latest claim.

Indeed, didn't he point out back in that Aug 2017 post that the newly-defeated National Opposition surged by a significant 7 percentage points in the polls following Jim McLay's toppling of Rob Muldoon in 1984 (at more or less the same point into the first term as Bridges' replacement of English) ? (2)

Bridges, of course, has achieved nothing of the sort for National. The Party has largely flatlined throughout the first half of 2018. But the point is: Farrar's latest contention - that it's nothing short of "astonishing" that National has managed to avoid a fall in poll support despite its leadership change - has been eloquently undermined by none other than the more youthful (and presumably wiser) Farrar of August 2017. In other words, he appears to have hoisted himself on his own petard. Thus doing himself a serious (reputational) injury.

But it's David Farrar's final three claims - centering on the notion that Labour has failed to receive some sort of traditional Budget Bounce in the Polls - that I want to highlight and test here.

These are his three contentions:
(1) There's usually a 2% bump (in Newshub polls) for the Government after a budget. But    Labour has flatlined.

(2) That, traditionally, New Governments reach "the height of their popularity" in the Polls carried out immediately following their First Budget, the Ardern Govt apparently proving an exception.

(3) That the percentage point gap - or "comfort margin" as Farrar calls it - currently enjoyed by the Ardern Govt over the Opposition is much narrower than it was for previous New Governments in the polls carried out immediately following their First Budget.

Testing: One, Two, Three

 (1) There's usually a 2% bump (in Newshub polls) for the Government after a budget. But Labour has flatlined.

A couple of things to note before we go on.

First, Farrar apparently considers sampling error unworthy of mention. Which is surely a little odd for National's official pollster, given that a "2% bump" is well within margin of error territory for a Major Party sitting on more than 40% support in a poll with 1000 respondents. However, because I want to test Farrar's claim entirely on its own terms, for the purposes of this analysis I'll adopt his tacit assumption (also held by far too many journalists) that opinion poll measures are razor sharp in their precision.

Second, Farrar's terminology here is quite sloppy. Initially he talks about a traditional 2 point bounce for "the Government" but then immediately focuses solely on Labour's current trajectory. I've noticed this imprecision - this clumsy conflation of the Government and the Main Governing Party - in some of his previous posts. So, just to be clear, we're dealing with a claim that isn't so much centred around the post-Budget trajectory of Government Blocs as a whole but rather solely that of the Major Party of Govt, whether it be Ardern's Labour now or the Clark Labour and Key National Parties of the recent past.

Third, a whiff of conceptual incoherence arguably pervades Farrar's claim. He obviously wants to seed the idea that budgets engender a spike in Major Governing Party support, with Ardern's Labour allegedly proving an exception. But the very fact that he is forced to significantly qualify by specifying a particular Pollster immediately casts all that into severe doubt. Do Governing Parties not enjoy a post-Budget bounce in the Colmar Bruntons and Roy Morgans, then ? And if they don't then what happens if Labour falls short of historic precedent in the TV3 poll but exceeds expectations in the Colmar Brunton ? I mean, precisely what electoral significance are we expected to extract from all that ?

Regardless of any latent incoherence, however, right-wing commentator Matthew Hooton was more than happy to dutifully regurgitate Farrar's claim (albeit without attribution) a few hours later on his Nine to Noon Politics segment (here):
I would have expected more than 0.3 per cent lift from Labour. In previous budgets, the governing party's gone up about 2 per cent in this particular poll after budgets (sic)

Well perhaps not all that dutiful because whereas Farrar had added a caveat - "There's usually a 2% bump" - you'll notice Hooton inadvertently enters no such qualification. It appears that, for Hooton, there is always a 2 point post-Budget bounce in TV3 polls.

To methodically test these two slight Farrar / Hooton variants of essentially the same claim, I've gone back through all of TV3's post-Budget polls since the year 2000 and, in each case, calculated support movement for the Major Governing Party. I've divided these post-Budget polls into those where Governing Party support fell, those where it rose and those where it remained steady.

If Hooton's right then we should not only expect every single year to be located in the Rise column but also for each and every increase in support to be more or less around the 2 percentage point mark (+2.0).

If Farrar's correct and there is "usually a 2% bump" then we should expect a substantial majority of the last 18 years to be located in the Rise column and to be of a 2 point magnitude, with the relatively small minority of remaining years presumably being distributed among the Fall and Steady columns.

Alas, even the most cursory of glances at Table 1 is enough to convey just how utterly groundless the Farrar-Hooton claim is.


Table 1: Major Party of Govt - Performance in
 Post-Budget TV3 Poll

       Fall                      Rise                 Steady

2016   - 1.9            2017   + 0.3           2007   = 
2015   - 3.4            2014   + 4.4           2002   =
2013   - 2.3            2010   + 3.2       
2012   - 4.0            2009   + 2.1
2011   - 4.5            2006   + 4.0
2008   - 3.0            2004   + 1.0
2005   - 5.0
2003   - 1.0
2001   - 1.0
2000   - 7.0


What we, in fact, find is that Major Parties of Government neither "usually" (Farrar) nor "always" (Hooton) enjoy a 2% bump in Newshub polls immediately following a Budget. Indeed, the clear tendency in TV3 polls since 2000 is for Budgets to engender a decline in support for the Governing Party. They have fallen after 10 budgets, risen following 6 and flatlined following 2.

What's more, a grand total of one solitary - I'll just repeat that, ONE SOLITARY - Budget over the past 18 years has generated a 2 point bounce (2009: +2.1).   Not all,  not most ... just one.

And even this one solitary "2% bump" in 2009 is not quite what Farrar / Hooton would have us believe. 

For one thing, National's modest bounce appears to have come almost entirely at the expense of its Minor Govt partners rather than the Opposition forces (Nat +2.1, Minor Govt partners - 2.3, Oppo +0.3).

For another, the 2009 post-Budget TV3 Poll was conducted well over two months after Budget Day, raising the possibility that other political events might have played a confounding role in the swing (such as it was).

With that last point in mind, let's zero-in on those specific years when TV3 were careful to conduct their Polls immediately before and after the Budget, thus ensuring that the popular response was measured with some precision, untainted by other political events. 

You can see (Table 2) that in these years things were even more lop-sided ... Seven falls in support for the Major Party of Government, ... just one rise, ... and absolutely zero / nyet / nil / nowt 2 point bumps.


Table 2: Major Party of Govt - Performance in
 Post-Budget TV3 Poll 

(Budget years where TV3 Polls conducted
 immediately before and after Budget - so
 not confounded by other events)

       Fall                      Rise                Steady

2013   - 2.3            2010   + 3.2       
2012   - 4.0
2011   - 4.5
2008   - 3.0
2005   - 5.0
2003   - 1.0
2001   - 1.0


Call it rubbing salt into the wound, but with Table 2 in mind, let me just remind you of the Farrar-Hooton assertion:
There's usually a 2% bump (in Newshub polls) for the Government after a budget. But Labour has flatlined.
In previous budgets, the governing party's gone up about 2 per cent in this particular poll after budgets (sic)

Notice from Table 1, incidentally, that the final (2017) Budget delivered by the Key-English Government engendered precisely the same modest +0.3 lift for National as that enjoyed by Ardern's Labour the following year. Did Farrar or Hooton immediately rush to publish a similar No Budget Bounce for National post, then ? Strangely enough ... No !  Hard to believe, isn't it ?  I mean, what are the odds that they'd craft that kind of narrative frame for one Govt but not the other ??? Certainly has me scratching my head.

We could, of course, adopt a remarkably generous interpretation of Farrar's claim and assume that he'd once again employed extremely sloppy language, thus inadvertently disguising his real intention.

Perhaps he was actually trying to suggest that New Governments after a First Budget received the "2% bump" in TV3 polls ? After all, the one solitary post-Budget 2 point bounce (+2.1) of the last 18 years was recorded following the first Budget of the incoming Key Govt in 2009.

It has to be said, though, this really is a rather generous interpretation. For one thing, Farrar's claim appears relatively straightforward:
There's usually a 2% bump (in Newshub polls) for the Government after a budget. But Labour has flatlined.
 No specific mention of a "New" Government or "First" Budget.

For another, if you look through Farrar's smattering of historic poll comparisons on Kiwiblog over the years you'll find that he's relied exclusively on TV1 Heylen / Colmar Brunton data for the pre-2002 period. I doubt very strongly, then, that he has any of the TV3 poll data gauging the popular response to the Clark-Cullen 2000 Budget, let alone the 1991 Bolger-Richardson Mother of all Budgets.

But being highly generous people, we'll give the young fellow the benefit of the doubt and assume (for the sake of a truly comprehensive scrutiny) that Farrar was, indeed, referring to TV3 polls following the First Budgets of New Governments.

While we're at it, we might as well also bring the other major Public Pollsters into the equation. Farrar does, after all, strongly imply in his final two claims that (regardless of Pollster) the initial Budgets of previous incoming governments did generate a clear (and perhaps hefty) poll bounce (hence his emphasis on these administrations reaching "the height of their popularity").

On the first point, the TV3 data in Table 3 shows that once again Farrar is entirely mistaken - even on the most generous interpretations of his "2% bump" theory.

Post-Budget TV3 polls actually recorded a significant slump in popular support for both the incoming Bolger Government (down 11 points) in 1991 and the incoming Clark Government (down 7 points) in 2000 (TV3 hadn't yet been established when the Lange-Douglas Govt delivered its first budget in late 1984).

We can confirm then that the Key Government was indeed unique in enjoying a 2 point post-First Budget bounce in the TV3 poll. Thus, regardless of interpretation, there is no Newshub 2% bump tradition in any shape or form. Farrar appears to have just made it up on the assumption that no one would bother checking.


Table 3: Major Party of Govt - Performance in
all Post-Budget Polls - First Budget of New Govt 
(Following Last 5 Changes of Govt)

                                        Fall          Rise         Steady

2018  New Ardern Govt
TV3 Reid Research                            + 0.3
TV1 Colmar Brunton                                             =

2009  New Key Govt
TV3 Reid Research                            + 2.1
TV1 Colmar Brunton          - 1.0
Roy Morgan                                                          =

2000  New Clark Govt
TV3 CM Research               - 7.0
TV1 Colmar Brunton           - 1.0
Herald-Digi                        - 7.6
NBR-UMR                           - 5.0

1991  New Bolger Govt
TV3 Gallup                        - 11.0
TV1 Heylen                        - 3.0

1984  New Lange Govt
TV1 Heylen                        - 3.0
Herald NRB                        - 4.0


On the broader point - Farrar's implicit suggestion that (regardless of Polling company) the first budget of previous new governments had always engendered a surge in popular support ... the answer would, once again, appear to be an unequivocal: No, not even remotely.

Table 3 clearly shows that, regardless of pollster, the Lange, Bolger and Clark Governments all lost support in polls conducted in the immediate wake of their first budget. In contrast, both the Key and Ardern administrations stand out for the neutral-to-mildly-positive response they received from the electorate.

Probably the most effective way to encapsulate all this for comparative purposes is to average the various pollsters' results for each New Government (Table 4).


Table 4: Major Party of Govt - Average Performance in
all Post-Budget Polls - First Budget of New Govt 
(Following Last 5 Changes of Govt)

                                        Fall          Rise         Steady

2018  New Ardern Govt                    + 0.15

2009  New Key Govt                         + 0.55

2000  New Clark Govt      - 5.15

1991  New Bolger Govt     - 7.0

1984  New Lange Govt      - 3.5


Pretty much says it all. The Ardern and Key Governments essentially flatlining, the three earlier governments falling. In other words, the public response to Ardern-Robertson's first budget is by no means the uniquely poor performance that Farrar would have us believe. Well above average would be a better description.

(2) That, traditionally, New Governments reach "the height of their popularity" in the Polls carried out immediately following their First Budget, the Ardern Government apparently proving an exception.

In a somewhat ill-tempered tete a tete with left-leaning bon vivant Psycho Milt in the Kiwiblog comments section, Farrar confidently repeated the assertion:
You really are being dense if you think that is all that matters. As I said, normally your first Budget is a high weather mark for a Government.
The notion Farrar wants to seed, here, is that New Governments are usually enjoying a popularity high in the immediate wake of their First Budget, whereas the Ardern Government is apparently heading in the opposite direction, sinking ever lower in public esteem (uniquely unpopular, unusually precarious, engendering "buyer's remorse" and so on and so forth).

If Farrar is right about the support trajectory of previous New Governments then we should, of course, expect (1) the Poll in which the particular New Government reached its height of popularity and (2) the post-Budget Poll to be one and the same thing. 

What we find, however, (Table 5) is that this was, in fact, never the case.  Regardless of Pollster, support for each New Govt reaches its apex in the immediate wake of the Election (although the precise timing naturally enough varies) and has then fallen (heavily in 1991 and 2000) by the time of the first post-Budget Poll. (3)
Table 5: Height of Popularity Poll Vs Post-Budget Poll 
for last Five New Governments

Post-1984                    Lab Govt               Nat Oppo

(1) TV1 Heylen

Height of Popularity
Late Sep 1984 Poll          48.4                       35.9

Post-Budget Poll
Dec 1984                         44.4                       42.1

Diff                                  - 4.0                     + 6.2

(2) Herald National Research Bureau

Height of Popularity
Oct 1984 Poll                  48.0                       33.0

Post-Budget Poll
Dec 1984                         44.0                       39.0

Diff                                  - 4.0                     + 6.0

Post-1990                   Nat Govt       Lab      Lab+Alliance Oppo

(1) TV1 Heylen

Height of Popularity
Dec 1990 Poll                  50.0          32.0            49.0

Post-Budget Poll
Aug 1991                          34.0          43.0            63.0

Diff                                 - 16.0        + 11.0        + 14.0 __________________________________________________________

Post-1999                    Lab+Alliance Govt    Nat+ACT Oppo

(1) TV3 CM Research

Height of Popularity
April 2000 Poll                      56.0                      32.0

Post-Budget Poll
June 2000                              50.0                      38.0

Diff                                       - 6.0                     + 6.0

(2) TV1 Colmar Brunton

Height of Popularity
March 2000 Poll                     56.0                    35.0

Post-Budget Poll
June 2000                               48.0                    44.0

Diff                                        - 8.0                   + 9.0

(3) Herald Digi

Height of Popularity
March 2000 Poll                      58.3                    30.1

Post-Budget Poll
June 2000                                 49.9                   38.9

Diff                                          - 8.4                  + 8.8
Post-2008                    Nat-led Govt    Lab+Green+NZF Oppo

(1) TV3 Reid Research

Height of Popularity
Feb 2009 Poll                      63.5                 35.6

Post-Budget Poll
Aug 2009                             60.9                 37.7

Diff                                      - 2.6                + 2.1

(2) TV1 Colmar Brunton

Height of Popularity
Feb 2009 Poll                      62.5                 35.4

Post-Budget Poll
July 2009                             60.1                 39.6

Diff                                      - 2.4                 + 4.2

(3) Roy Morgan

Height of Popularity
Late Feb 2009 Poll              61.3                 37.5

Post-Budget Poll
Early June 2009                   57.5                 42.0

Diff                                      - 3.8                 + 4.5


Post-2017                   Lab-led Govt      Nat-ACT Oppo

(1) TV3 Reid Research

Height of Popularity
Jan 2018 Poll                      52.1                    44.7

Post-Budget Poll
May 2018                            50.7                    45.1

Diff                                     - 1.4                    + 0.4

(2) TV1 Colmar Brunton

Height of Popularity
Feb 2018 Poll                      55.6                   43.5

Post-Budget Poll
May 2018                             52.2                   45.7

Diff                                      - 3.4                   + 2.2


Moreover if you average the various Pollsters' results for each New Government (Table 6) ... lo and behold ! ... you find that, far from suffering an unusually severe downward slide, the Ardern Govt's current poll rating is actually closer to its height of popularity than is the case for any of the previous four incoming governments.

The post-Budget Ardern Govt is down a mere 2.4 points from its apex, compared with the Key Govt's 2.9 point slide, the Lange Govt's 4 point fall, Clark's significant 7.5 point slump and Bolger's entirely woeful 16 point plunge.

Table 6: Height of Popularity Poll Vs Post-Budget Poll 
for last Five New Governments (Average of all Polls)

Post-1984                    Lab Govt               Nat Oppo

Height of Popularity        48.2                        34.5

Post-Budget                     44.2                        40.6

Diff                                  - 4.0                      + 6.1

Post-1990                   Nat Govt       Lab      Lab+Alliance Oppo

Height of Popularity       50.0            32.0              49.0

Post-Budget                     34.0            43.0              63.0

Diff                                - 16.0         + 11.0           + 14.0


Post-1999                    Lab+Alliance Govt    Nat+ACT Oppo

Height of Popularity                 56.8                        32.4

Post-Budget                               49.3                        40.3

Diff                                           - 7.5                       + 7.9 

Post-2008                    Nat-led Govt    Lab+Green+NZF Oppo

Height of Popularity            62.4                     36.2

Post-Budget                         59.5                     39.8

Diff                                      - 2.9                    + 3.6


Post-2017                     Lab-led Govt       Nat-ACT Oppo

Height of Popularity            53.9                     44.1

Post-Budget                         51.5                     45.4

Diff                                      - 2.4                    + 1.3


In other words, the Ardern Govt's support is holding up unusually well ... the previous four new governments having all fallen further in their Post-First Budget polls relative to their Height of Popularity a few months earlier. Pretty much the polar opposite of Farrar's main thrust here.

(3) That the percentage point gap - or "comfort margin" as Farrar calls it - currently enjoyed by the Ardern Govt over the Opposition is much narrower than it was for previous New Governments in the polls carried out immediately following their First Budget.

Before considering the "comfort margin", let's just briefly look at the core components from which that margin is derived (Table 7). They comprise important measures in their own right.

First, the Opposition's overall share of support. We're regularly informed by Farrar, Hooton, Hosking, Trotter, du Plessis-Allan and others that the current Opposition is extremely - indeed uniquely - popular. In reality, as Table 7 shows, the 2018 Bridges-led Nat-ACT Opposition (45.7%) is a pretty poor second to the Labour-Alliance Opposition of 1991 (63.0%) and only a whisker ahead of the Shipley-led Nat-ACT Opposition of 2000 (44.0%).

Indeed, if you ignore questions of close ideological proximity and compare oppositions in their widest possible sense, then you'd not only see the June 2000 Nat-ACT-NZF Opposition on 45.0% (thus even more closely breathing down the current Opposition's neck) but you'd also have a combined National-Social Credit Opposition in Dec 1984 on 47.7% ... relegating the 2018 Bridges Opposition to third out of five and thus merely average.

That, in turn, immediately raises a crucial question about the (all too often ignored) other side of the coin - the Ardern Govt's comparative overall support share in historical context. As you can see, their 52.2% post-Budget rating (Table 7) puts them firmly in second place. Obviously nowhere near the extraordinarily good performance of the Key Govt in its infancy (60.1%) but still a pretty comfortable Number 2. 

Table 7: Government "Comfort Margin" over Opposition 
in First post-Budget Poll of Last 5 New Govts 
(TV1 Heylen / Colmar Brunton)

Lange Govt 1984     Lab Govt         Nat Oppo          Lead

Post-Budget Poll
Dec 1984                     44.4                  42.1           Govt  +2.3


Bolger Govt 1990     Nat Govt    Lab+All Oppo      Lead

Post-Budget Poll
Aug 1991                      34.0                63.0            Oppo  + 29.0


Clark Govt 1999  Lab+All Govt    Nat+ACT Oppo      Lead

Post-Budget Poll
June 2000                     48.0                 44.0             Govt + 4.0


Key Govt 2008     Nat-led Govt     Lab+G+NZF Oppo    Lead

Post-Budget Poll
July 2009                      60.1                    39.6          Govt + 20.5


Ardern Govt 2017    Lab-led Govt     Nat-ACT Oppo      Lead

Post-Budget Poll
May 2018                       52.2                     45.7            Govt  + 6.5


None of which augurs particularly well for Farrar's core contention here: that the Ardern Govt's "comfort margin" over the Nat-ACT Opposition is unusually narrow following their first budget.

And, as Table 7 confirms, Farrar is indeed entirely wrong. Far from being a very poor last as he would have you believe, the Ardern Administration's comfort margin over the Bridges-led Opposition (+ 6.5 points) is, in fact, the second widest of the last five incoming governments. A solid, above average performance. (4)

Along with this accumulation of empirical evidence directly contradicting Farrar's cluster of claims we might also (as a kind of denouement to this analysis) briefly explore one further illuminating measure that he chooses to evade altogether.

Incoming governments win elections with varying comfort margins - or percentage point leads - over their respective Oppositions. And it's widely agreed that the 2017 Election was particularly closely-contested, with the Ardern Govt's margin narrower than usual.

Arguably, then, the exact size of the post-Budget margin isn't quite so important. Instead, the more telling dimension - the one that reveals most about the longevity of a new administration's honeymoon - is the extent to which they've managed to maintain or expand that election night margin by the time of their first Budget.

Table 8 shows that on this measure, the Ardern Government's performance is yet again solidly above average.

The Election night lead enjoyed by the Lange and Clark Labour Governments had more than halved (down 5 points in both cases) by the time of their first post-Budget polls, while the Bolger Govt's margin had been entirely decimated and transformed into a significant deficit. (5)

By contrast, the Ardern Administration's gentle 1 point expansion was overshadowed only by the extraordinary tripling of the Key Govt's comfort margin (up by more than 14 points).

Table 8: Election Result Lead  Vs  Post-Budget Lead 
for Last Five New Governments (TV1 Heylen / Colmar Brunton)

                             1984 Lange Govt            1990 Bolger Govt        

Election Lead             Govt 7.1                          Govt 0.6        
Post-Budget Lead       Govt 2.3                         Oppo 29.0

Diff                                - 4.8                                - 29.6

                              1999 Clark Govt             2008 Key Govt

Election Lead              Govt 9.0                         Govt 6.1
Post-Budget Lead        Govt 4.0                        Govt 20.5
Diff                                 - 5.0                               + 14.4

                                           2017 Ardern Govt

Election Lead                           Govt 5.5
Post-Budget Lead                     Govt 6.5

Diff                                             + 1.0

Again, on a broader note, one might mention that Farrar's implicitly positive portrayal of the support trajectories and comfort margins of previous incoming governments is at odds with data he himself sets out in one or two Kiwiblog posts from previous years. See for example: his 2009 Post election polls for new Governments


Between his gross misrepresentations and glaring internal contradictions, National Party Pollster David Farrar has managed to fundamentally subtract from our understanding of the Ardern Government's comparative popularity.

Indeed, in a veritable negative tour de force, he manages to get every single facet wrong.

Seeking to reconcile politically-motivated spin with radically incompatible empirical data, Farrar is forced to brazenly conjure up historic opinion poll trends out of thin air, asserting the absolute antithesis of historical reality, while concurrently finding himself mired in embarrassing internal contradictions as old Kiwiblog posts come back to haunt him.

The picture Farrar paints is decidedly grim. Despite being so early in its first term, a uniquely unpopular and precarious Ardern Administration finds itself down on its luck and unable to catch an even break. Nothing more than a Coalition of Losers ... on the back foot right from the very start.

Previous incoming governments, according to the National Party maestro, had soared to unprecedented heights of popularity, reaching the broad and sunny uplands of success, after delivering their first budgets. As a result, he strongly implies, they commanded a very comfortable lead in public opinion over the Opposition forces of the day.

In stark contrast, Farrar would have us believe, the poll performance of the current Labour-led Government is remarkably poor. Stuck with a surprisingly razor-thin advantage over the opposition in the immediate wake of its first Budget, the Ardern Govt's support is either ebbing or flatlining when historic precedent says it should be surging and soaring.

Why, it couldn't even manage to achieve the "2% bump" in popular support that any government worth its salt would normally expect their budgets to engender !  I mean, what kind of an amateurish bunch of clueless no-hopers are they ???

Heading on back to reality for a moment (strap yourself in, David, this could be a bumpy ride) ... of the last five incoming governments, the Ardern Administration is, in fact, second only to its Key-National predecessor in (i) its overall support share, (ii) the size of the comfort margin it enjoys over the Opposition, and (iii) the extent to which it has managed to expand that margin since the Election.

Moreover far from suffering "a record short honeymoon" (see Farrar tweet footnote (6)), the Ardern Govt's honeymoon bounce is holding up unusually well - the four previous new governments (including Key's National in 2009) having all fallen further between their height of popularity and the immediate wake of their First Budget a few months later. The 2017 General Election may have been unusually closely-contested but those voting decisions have also remained unusually stable throughout the post-Election period. Significantly less opinion poll support volatility than in the past.

Meanwhile Labour's modest lift in the 2018 post-Budget TV3 poll (+0.3) and post-Budget all-poll average (+0.15) represents an above average performance regardless of interpretative lens or comparative criteria - the TV3 "2% bump" tradition being yet another Farrar foray into the exciting world of creative fiction - a genre in which he would appear to possess some latent talent.

Overall then, a pretty solid - if not exactly spectacular - performance by the Ardern Administration ... one they're entitled to feel quietly pleased with. Nowhere near, of course, as remarkable as the Key National Government's absolutely stellar post-2008 ratings but without a doubt well above the average for new governments over the past 4 decades. (7)


(1) In other words, with National vs Labour framing you're comparing virtually the entire poll support for the political Right (given that Right voters coalesce almost entirely around National) with poll support for just one segment (albeit the largest segment) of the political Left (given that the latter voting-base is much more fragmented among various parties).

This kind of sleight-of-hand might be permissible for known partisans like Farrar, Hooton and Hosking ... but it's extraordinary that certain mainstream journalists continue to get away with it.

(2) Characteristically sloppy language from Farrar in this August 2017 post: What is the Normal Poll Boost for a New Leader ?  You can't help but develop the distinct impression from some of his opening clauses that Farrar is suggesting New Major Party Leaders traditionally receive a boost to their own Preferred PM ratings.

For example:
New leaders almost always get a boost in the polls ...
I've gone back through all the One News polls since 1974 to look at what has been the poll boost has been (sic) for a new leader

It's only as you read on and scan the data that it begins to dawn that he's talking about new leaders boosting their Party's poll ratings.

Notice too that, oddly enough, Farrar conspicuously fails to follow his own explicitly stated criteria in this August 2017 post. He tells his readers that he is:
excepting those (new leaders elected) after an election loss

If  by "election loss" he's referring to a newly-defeated major Opposition Party that has just been turfed out at the previous election - then what, pray tell, are Jim McLay (elected leader just a few months after National's July 1984 defeat), Jim Bolger (replacing McLay still during National's first term in opposition) and Bill English (toppling Jenny Shipley in 2001 during the Nats' first spell in opposition) all doing on Farrar's list ?

Indeed, it looks to me like Labour's Phil Goff is the only new leader "after an election loss" that Farrar excludes.

If, on the other hand, Farrar more broadly means new leaders elected after any election loss (in other words, all newly elected Opposition leaders regardless of first, second or third term), then - along with McLay, Bolger and English - ... Lange, Clark, Brash and Key shouldn't be on the list either. I mean, what the flying hell ???

(3) A couple of things to clarify regarding the data in Table 5.

First, note that support for both the Lange and Clark Governments reached new peaks of popularity much later on (albeit with one specific Pollster in each case - TV1 Heylen for Lange / TV3 CM Research for Clark).

As Table 5 shows, the Lange Administration reached an initial apex of 48.4% a couple of months after its July 1984 Election victory (TV1 Heylen). It then fell by 4 points to 44.4% in the December 1984 post-Budget poll (the one Farrar wrongly claims recorded the Govt's peak of support for its entire first term).

But then from mid-1985 (around one year after the Election and more than 8 months following the First Budget), the Lange Govt began to reach new heights of popularity, largely fluctuating over the next 2 years between the mid-40s and mid-50s. By the onset of the 1987 Election campaign, however, it was starting to nudge a very impressive 60%.

The Clark Govt reached an apex of 56.0% in both the TV1 Colmar Brunton (March 2000) and TV3 CM Research (April 2000) (Table 5). That remained their height of popularity in the Colmar Bruntons (though they came very close again in June 2002 as the Election campaign began to heat up). But in the three TV3 CM Research polls conducted between February and June 2002, they exceeded that rating, scoring 56.5% (Feb), 59.4% (April) and 57.2% (June).

In order to keep things simple (and also because we're comparing the support trajectories of these earlier incoming Govts with those of an Ardern Govt that is less than a year old), I've therefore focussed on each new Govt's initial peak during its first year in Office. If, instead, I'd compared the post-First Budget poll support with these much later peaks in popularity then that would, of course, demolish Farrar's claim in an even more emphatic manner.

For example, June 2000 Post-First Budget Clark Govt support would be down not a mere 6 points (as it was compared to the initial April 2000 apex) (TV3 CM Research) but, in fact, down by a more significant 9.4 points (from the much later April 2002 apex).

The second thing to clarify: all 2009 TV1 Colmar Brunton figures presented in Table 5 (and, indeed, in various other Tables in this post) are calculated from the official Colmar Brunton Poll report of September 2009 (see Upper screenshot immediately below). I mention this because (at the time of publishing this post) the relevant Colmar Brunton figures listed on the appropriate Wikipedia page (see Lower screenshot below) are incorrect.

To show where Wikipedia went wrong, I'll focus solely on the Nat / Lab / Green stats (for simplicity).

         Nat / Lab / Green %
Colmar Brunton (Early-Mid 2009)

                    14-19 Feb      28 March-1 April       5 April         25-29 July

CB Report     56 28 6               57 31 7                    X              56 31 7 
Wikipedia      57 31 7               56 28 6                57 31 7         56 31 7

Clearly, Wikipedia's July 2009 (Post-Budget) CB figures are correct, but figures from the earlier part of the year are wrong.

The Feb and March stats are quite obviously erroneously transposed - the mixup probably occurring because Wikipedia's source material was a TVNZ teletext that mentioned the results of both polls together. Meanwhile the (1) 28 March-1 April and (2) 5 April Colmar Bruntons are, in fact, one and the same Poll (the former are the fieldwork dates and the latter the release date for the poll).

(4) As you can see, I've extracted Farrar's final two claims from the last paragraph of his No budget bounce for Labour post. Here's the relevant passage:
Now of course the centre-left still has more support and seats than the centre-right. But their comfort margin is razor thin at a time when they should be at the height of their popularity having delivered their first Budget

I think it's reasonable to infer two distinct claims from this passage:

(i) The Ardern Govt's "razor thin" post-First Budget comfort margin is narrower than is traditionally the case for incoming governments.

(ii) New administrations traditionally reach the height of their popularity immediately following their first Budget, with the incoming Ardern Govt an exception.

That, of course, still leaves the issue of precisely what Farrar means by the designation: "centre-left".

As far as I'm concerned, detailed NZ Election Study findings over the last decade have amply demonstrated that NZF is very much part and parcel of the broader Centre-Left. The Party's voting-base is best seen as occupying the morally conservative end of the economically Centre-Left spectrum.

On the face of it, of course, it's quite possible Farrar was referring solely to the Labour+Green Bloc. Had that indeed been his intention then it would simply have constituted yet another deliberate sleight-of-hand on his part - casually lobbing off a whole chunk of the Ardern Government's support-base (ie NZF supporters) in order to make its opinion poll lead look more precarious than it actually is. Like his FPP-style National vs Labour framing, such an action would seem little more than manipulating the data and shifting the goalposts in the service of political spin.

Fortunately, however, Farrar subsequently clarified what he meant by "centre-left" in his brittle reply to Psycho Milt:
You really are being dense if you think that is all that matters. As I said, normally your first Budget is a high weather mark for a Government ... and as NZF remain low, their normal instability will strike.

Hence, I've taken it that his final two claims refer to the Ardern Government's post-First Budget poll ratings vis-a-vis those of its recent New Government predecessors.

I assume, incidentally, that Farrar meant "high water mark" rather than "high weather mark". He'd presumably become momentarily confused with an entirely different phenomena: highly marginal weather vane seats that pundits traditionally focus on in FPP elections because they tend to go with whoever wins government at any particular election.

(5) The exact composition of Opposition forces in 1991 is a little more ambiguous than usual when it comes to calculating poll support.

On the one hand, at the beginning of 1991: the two Opposition parties represented in Parliament were Labour and New Labour (the latter represented by Jim Anderton in his Sydenham seat).

On the other hand, it was well-known by the time of the first post-Mother of all Budgets poll that the Greens had agreed to join New Labour in a new Left-leaning political vehicle: The Alliance ultimately forming toward the end of 1991. With Anderton as leader, the Alliance would, of course, already be represented in Parliament.

The crucial question then is: Do you (1) restrict Opposition poll support solely to Lab + New Lab up until the formal launch of the Alliance ? or do you (2) also add Green poll support from earlier on in 1991 - on the basis that prospective Green voters already knew that the Party they were supporting was about to join the official Opposition ?

In the various tables of this post, I've chosen the second of these two options - not least because, back in 1991, there was already a tradition of adding Green support to the overall Opposition figure. For example, well before the Alliance was officially launched in late 1991, various newspaper opinion poll reports were already aggregating New Labour and Green support in anticipation of the new Party. Furthermore, the 1993 New Zealand Election Study took the same route for the entire period, aggregating the early-mid 1991 poll support for the various parties that subsequently went on to form the Alliance.

So the obvious question is: what happens if you instead follow Option (1) and subtract the Greens from 1991 Opposition poll support in the various tables ? Does this in any way undermine the core conclusions of my analysis ?  Does it weaken the demolition of Farrar's various claims ?

The answer is an unequivocal: No

Let's take the analysis surrounding Tables 7 and 8 as examples.

For Table 7, what happens if we lob the Greens off the 1991 Oppo figures ? Post-Budget Oppo poll support falls from 63.0% (Lab+NL+Green) to 52.0% (Lab+NL) with the Opposition lead over the Bolger Government falling from + 29.0 to + 18.0. None of which makes any appreciable difference to the core findings of my analysis - the Ardern Govt's comfort margin over the current Opposition continues to be second widest (as opposed to Farrar's claim that it was the narrowest of the last five incoming governments). And overall post-Budget support for the current Bridges-led Nat-ACT Oppo (45.7%) is still a pretty poor second to the Lab-New Lab Oppo of 1991 (52.0%) - albeit, of course, the gap nowhere near as emphatic as it is if you include the Greens in the 1991 Oppo (63.0%) (Thus demolishing claims by Farrar / Hooton / Hosking and others that the current Oppo is extremely / uniquely popular).

For Table 8, subtracting the Greens sees the 1990 Bolger Govt's Election lead increase from Govt by 0.6 points (over Lab+NL+Green) to Govt by 7.5 (over Lab+NL) with the post Budget Oppo lead declining from 29.0 to 18.0. The lead thus swinging away from the Bolger Govt by 25.5 points (with Greens excluded) rather than 29.6 (with Greens included). None of which, of course, in any way alters the core conclusion for Table 8 that the Ardern Govt's one point expansion in support is solidly above average.

(6) Farrar was referring specifically to Labour's poll performance in this particular tweet - but it clearly also captures the spirit of his Kiwiblog argument that previous incoming governments were reaching their honeymoon heights after their First Budget, while the Ardern Govt was supposedly down on its luck ... its margin razor thin.


(7) None of which, of course, is to imply that the Ardern Govt has any cause to rest on its laurels.

For one thing, you have the obvious concern that Labour's junior partners may potentially fall short of the 5% hurdle at the 2020 Election. A 6.5 point Government advantage over the Opposition won't amount to a hill of beans under that scenario.

For another, they'd do well to remember that their historic (incoming govt) competition wasn't necessarily all that flash at this point into the first term. The Clark Govt, for instance, was already descending into its Winter of Discontent as it delivered its first Budget in 2000. And, although now largely forgotten, Ruth Richardson's 1991 Mother of all Budgets immediately provoked the Mother of all Backlashes from the New Zealand public, as the poll data so strikingly reveals.

So there are sound reasons for avoiding complacency.

But then, all of that just highlights once again how fundamentally Farrar has obliterated the historical record. Pressed into his conceptual meat grinder, even Clark and Bolger's respective plunges and freefalls are radically recast as fresh new governments soaring to unprecedented heights.

Given this blatant misrepresentation of public opinion, you might ask why, at the outset of this post, I essentially dismissed Farrar's latest cluster of claims as merely a "faint echo" of his December 2017 Honeymoon Scam ?

The answer lies in the contrasting degrees of influence he's been able to exert across the media. Pleasingly enough, journalists and commentators this time have displayed far greater scepticism (or, at least, caution) over his claims, with only Hooton and Clifton happy to take the bait with open arms.

It's true that he may have managed to exert a somewhat more subtle influence on other segments of the media. One or two news media reports, for instance, emphasized that Labour had failed to receive a boost from the Budget (eg Greens almost out in new poll, Budget does not give Labour a boost (Stuff) , 1 News Colmar Brunton Poll: Budget fails to deliver boost for Labour (TVNZ) ). But the evidence that Farrar had any influence here is, shall we say, somewhat oblique. And purely speculative.

Call me a vainglorious bastard if you will, but I like to think my earlier critique of Farrar's scam has played at least some role in the media's greater caution. I originally published Farrar's Honeymoon Scam on January 19 and advertised it on The Standard  the next morning. Over the following two weeks, readership slowly climbed to around 950 pageviews (my previous most popular post had attracted just over 1200). And then on February 9 ... All Hell Broke Loose ! First, leading Economist and Social Scientist, Brian Easton, posted a typically astute (and very generous) overview of my critique, before going on to explore some of the wider implications. (here)

That led, in turn, to some very high profile activity on Facebook. Freelance editor and writer, Claire Gummer, linked to the post on Facebook's Kiwi Journalists Association page (here), prefacing with the question:   Do the news media rely too heavily on David Farrar ?  This not only generated some debate among journalists and others but also inspired Tim Roxborogh to link on the popular  Newstalk ZB  'The Two'  Facebook page (here), again with a very positive summary.  Reddit (here), Pete George at Your NZ (here) and one or two other forums subsequently discussed the issue over the next few days, while Left-leaning commenters linked to the Post on both The Daily Blog (here) and (somewhat mischievously) on Kiwiblog itself (here), with various others tweeting the link with generous comments.

(Surprisingly enough, some of the Kiwiblog commenters weren't quite so fulsome in their praise ... but they did give me a bit of a laugh (albeit unintentionally so). The broad consensus among the Kiwiblog cognoscenti was that I was almost certainly a hardcore Left-wing academic, although one dissenting opinion suggested that only a deeply unhinged personage could possibly be so gauche as to critique Farrar's modus operandi).

All of which generated the sort of explosion of interest that I usually associate with the Big Three Political Blogs: currently the post has received just a little over 11,200 pageviews (largely from Facebook). For a relatively obscure (and let's face it, at times almost defunct) blog like this one - those sort of numbers are Space Age stuff !, absolutely futuristic !, almost like living in the year 1999 !!!

A further indication that my critique may just have tip-toed to the outer edges of the media zeitgeist for a day or two back in February was this oblique reference in the final sentence of a Dominion Post Editorial:

Editorial: National MPs are fighting over a poisoned chalice (Feb 21 2018):
The argument by National propagandists that Labour failed to get a post-election bounce has now been disproved. Labour's bounce is huge

Hard to be sure, but I thought the term "National propagandists" was a bit of a giveaway here. But, then again, maybe I'm just a total narcissist with delusions of grandeur. I suspect at least one Kiwiblog regular would wholeheartedly agree. 😀