After my post yesterday I remembered that I had also written a post in 2019 based around the excellent talk Jenny Wilkinson, then the Australian federal Parliamentary Budget Officer had given at Treasury in 2019. I ended that post this way

….it was a very useful presentation (I hope Treasury makes her slides available) from a technocrat’s technocrat.  I’m left sceptical on two main counts:

• first, whether elections ever much do, or really should, turn much on precise fiscal costings. Perhaps it appeals to inside-the-Beltway technocrats to conceive of that model, but I see elections as mostly about things like competing visions, competing personalities, competing diagnoses, and competing claims to competence.  If so, why spend so much on highly-detailed and expensive state-funded costings, that the parties themselves don’t think it worth spending their own money on?
• second, we should think harder about the whole panoply of support and information etc we provide to political parties and the public, preferably without further reinforcing the favoured position of established large parties.  Thus, it is interesting to note that written parliamentary questions are much much less used in Australia, as a way of garnering information, than is the case in New Zealand. (“In the years 2008–2014 only about 8 questions in writing were being asked each sitting day, but this number increased to 19 in 2015, and was 14 in 2016.”).   What about better resourcing select committees (to me a better use of money)?  And if we threw in a free PBO service, should we reduce existing money parliamentary parties are funded with?  If not, why not?  And would resistance to that idea suggest the costings were some epicurean nice-to-have rather than a central element of a well-functioning democracy?  And then, of course, there is the OIA.  Mightn’t it be better to require agencies to release documented costings models themselves, in ways that would allow political parties and their consultancy firms to use them to the extent they judge appropriate (and not otherwise).

And if I had the analytical resource implied by 40-45 more staff and had to deploy it somewhere in the public sector, it is far from obvious that a policy costing operation (with supporting analysis and research as the PBO) would offer the highest benefit-cost ratio.

Rereading that got me thinking again about the resource requirements for such an agency in New Zealand, that both Grant Robertson and Nicola Willis now seem keen on (despite apparently straitened fiscal circumstances). As I noted in yesterday’s post, no small advanced economy I’m aware of runs one of these costings offices (Australia, you will recall, has five times our population and rather more than that multiple of real GDP, the real resources used for this luxury product). And policy issues aren’t really less complex or less numerous just because your country is smaller (Australia has some federal/state interaction issues, but they aren’t likely to material affect the potential demand for policy costing work).

As far I can see there are three such costing offices in Australia. I will focus on the federal and Victorian versions, but the first such entity was the New South Wales one.

The NSW entity is a bit of an odd beast, and I don’t think anyone has championed anything like it in New Zealand. It is set up only for the 9 months prior to each state election (not sure if snap elections are allowed in NSW), and can be used only by the leaders of the two main parties, who in turn are required to submit all their policies to the PBO 10 days before the election, and the PBO is required to publish costings for them at least 5 days before the election. It seems to be staffed largely by temporary secondees from existing public service departments, overseen by an academic. From the report on the 2019 election, it seems to have had about 20 staff at peak

In practice, it seems that parties work with the PBO behind the scenes in advance getting their policies costed, and either modifying or dumping ones that come in too expensive etc, with only the final policies and the costings of them seeing the light of public day.

If you believe in these sorts of things, I guess one can see the logic of the NSW approach, as it tries to put the main Opposition party on something like the same footing as the governing party. It isn’t a small financial or resource commitment (about the total staff numbers of, say, our Productivity Commission), for what is after all only a state government, but it is only for 9 months every three years.

What of the federal Parliamentary Budget Office? There is quite a lot of material in that earlier post. One other extract that might be worth bringing forward is

…in answer to a question from me, Wilkinson observed that what the PBO can best do is cost programmes that represents small deviations from the status quo (they have good tools to estimate direct and immediate fiscal costs/gains) while wider economic second round effects, and the associated fiscal impacts, are likely to be small.  But, and using her own (deliberately extreme) example, if some party were to campaign on getting rid of the welfare state, her office could do the direct fiscal costs, but could offer little or nothing on the wider economic (or social) effects of such a policy, including the possibility that it might have large long-term indirect fiscal implications.    They will only offer qualitative statements about those wider effects.  Which left me thinking that the the PBO probably does very well on things that don’t matter that much, and can’t offer much on the bigger issues that elections probably should really be about (whether about the welfare state, climate change, productivity or whatever).   

They do a lot of costings

Another aspect of the presentation that surprised me was (a) the number of costings the PBO does, and (b) the extent to which demand is not concentrated just in the pre-election period.  In fairness, she noted that the latter had surprised them too.  In the most recent year (an election year) they’d done 2970 costings, while in the previous two non-election years they had averaged about 1700 costings. Only MPs can request costings, and there are 227 MPs (across House and Senate).     Those numbers don’t mean 2970 separate items of policy, as many of the costings will be, in effect, rework as members or parties iterate towards a policy that meets their ends and will be scored by the PBO as not costing too much.

but note (see above) how much less extensively written parliamentary questions are used in Australia.

At the time of that 2019 presentation, Wilkinson told us her office normally had about 45 staff, scaling up to around 55 at elections (and recall that this is a federal government, in a system where a lot of policies are state responsibilities). The table in the latest Annual Report suggests that is still about right

What of the Victorian state Parliamentary Budget Office? Here is what they say they do

All MPs have access (unlike in NSW).

This doesn’t come cheap. Their documents say that for last year’s election they peaked at 26 FTEs, and they appear to have a permanent staff of about 16.

For a system of unitary government it seems reasonable to think in terms of the combined Victoria plus federal offices (which thus cover all the policy areas that affect Victoria, whether via federal or state policy). That seems to involve a base level of 60 staff, scaling up to perhaps 80 in the run-up to elections.

Now, of course, Australia is a fairly big country. But as already noted, neither the number of policy issues nor the design complexity of those policies is really scalable with the size of the country. And it seems most unlikely that one could do a worthwhile job – across the multiplicity of areas of policy – with 12-16 staff in New Zealand. In fact, I find it difficult to see how it could be done well – and there is really no point if it is not done well – with fewer than perhaps 30-40 staff.

That would be bigger (much bigger) than either the Productivity Commission or the Parliamentary Commissioner for the Environment (21 staff). Would it be a priority use of scarce resources for taxpayers to be putting this additional financial assistance towards political parties and MPs? I continue to think that better-resourcing select committees would have a much larger payoff for citizens and good governance. There is likely to be a good reason why no other small advanced countries run state-funded policy costing offices (while parties themselves are of course free to use economics and other consultancy firms to the extent they find useful – in a political market).

Issues of scale are very real for small countries’ central government policy functions. I’ve already mentioned that the Productivity Commission has 15-20 people in total. To the extent there was an inspiration behind our Commission (as distinct from a bauble for the Key government to throw ACT’s way), it was the Australian Productivity Commission, which has over the years produced a lot of useful reports. The latest Annual Report suggests that Commission has 165 staff and 12 commissioners. The sorts of issues facing Australia are not likely to be any less numerous or complex than those facing New Zealand, and even if the staff of our Commission walked on water they simply could not match the value that could be added by the Australian Productivity Commission.

I would have thought there was no credible way we would devote 180 people to the Productivity Commission – nor do I think we should – but I guess if the Ministry for the Environment now has more than 1000 staff really who knows anymore. But with 15-20 people it was always going to be vulnerable to going the way it has, and was always going to struggle to maintain depth and critical mass.

The pool of really able people is small, and fiscal resources are limited (in a smallish underperforming economy). It just doesn’t make a lot of sense to be thinking of putting dozens of people into helping political parties cost their specific policies. On the track record of their performance, nor does having 15-20 people in a Productivity Commission that now seems, in practice, to perform a more general role in support of political parties of the left. A new government should rebuild Treasury, and if there is resource it would be better spent strengthening select committees to scrutinise actual legislation and actual government agencies, rather than (further) funding the bids of the parties competing for the keys to the Beehive.

Yesterday one of the Labour’s surrogates – Craig Renney, economist at the CTU, but also former adviser to the Minister of Finance (and reputed to be interested in being a Labour MP himself) – came out with a short document attempting to put a fiscal frame around National’s election promises. (One might have thought that if you’d been a part of enabling an $11bn hole in the government finances – the LSAP losses – you might have been a bit more modest in your rhetorical tone, but I guess he was only a (very senior) adviser).

It is explicitly partisan political in nature, and heavy on rhetoric. The flavour is perhaps conveyed by the front cover

As anything other than partisan spin though, it was a strange document. I’m as keen as anyone to see National’s programme (with numbers), but then in a week when the PM has been going round distinguishing between Labour Party policy and government policy, the same could be said of Labour’s (as yet largely unannounced) programme.

Renney’s document is built around an attempt to show that what National has announced so far does not fit within the operating allowances for the next three Budgets set down by the Minister of Finance in this year’s Budget. I have no reason to doubt that his numbers are approximately right. But (not only is it not clear that National has yet announced all its policies) there is no obvious reason why National would regard itself as bound by the operating allowances the Labour Minister of Finance had put in his pre-election budget. After all, Labour hasn’t in the past, and is highly unlikely to do so next year were it to be re-elected. (The operating allowance framework is in any case quite badly flawed as any sort of future signal, but especially when inflation is jumping around.)

I’m not championing what we know of either side’s fiscal policy. From both sides, there seems to be a disconcerting falling away from a commitment to budget surpluses, except in that vague distant future sense of the early St Augustine (“Lord, give me continence and chastity, but not yet”).

But Labour’s own (official government) numbers already have about them a considerable air of unrealism. In this year’s Budget, the only hard numbers – those planned for 23/24 – showed a slight increase in core Crown primary spending (ie excluding finance costs) as a share of GDP, but then the vapourware numbers – the ones relying on those operating allowances – show a fall from 31.2 per cent of GDP this year to 29.7 per cent in 2026/27. Another way of looking at those same numbers is to calculate – all from Budget numbers – real per capita core Crown primary spending over those three years Renney focuses on. On Labour government numbers, real per capita expenditure is projected/planned to show no growth at all in the next three years. Does that seem like a prospect that would align with what we’ve seen of this government’s approach to spending in recent years (in an era of ageing populations, public sector wage pressures etc)? Not to me. And the last three years have seen a single party government, and on all the polls if Labour were to get back in it would be dependent on the even less fiscally disciplined, less inclined to expenditure restraint, Green and Maori parties. With a Labour leader who has already ruled out the wealth taxes those two parties favour to help pay for their fiscal ambitions.

There is also the small matter of scale. Renney’s claims are that there is a gap of $3.3-$5.2 billion over the three fiscal years in questions. Since core Crown primary spending over those years is estimated at in excess of $400 billion any such gap is 1 per cent spending (or about 0.3 per cent of GDP). Seems a slim basis for such florid headlines.

But it will be good to see National’s numbers. In a better world, they would be credibly showing a commitment to a structural fiscal surplus next year. But given that Labour’s vapourware tiny surplus the following year was looking shaky from day 1 (once Eric Crampton pointed out the tobacco excise tax losses they and Treasury had accidentally left out) I don’t suppose it is likely.

But no doubt Renney’s report achieved its political end and put National and Willis on the back foot for a news cycle. National’s response didn’t seem much better. It seemed to consist first of suggesting that the economist of the CTU shouldn’t be commenting in public, which was a bit odd to say the least. And then we had the attempt to reclaim the news cycle by suggesting that it all (what?) was down to Grant Robertson not having followed through on the earlier plan to set up a budget-costing unit. National had (rightly in my view, see below) opposed the idea of such a unit when Labour and the Greens were championing it, but apparently when Willis had become finance spokesperson she had written to Robertson to express National now being in favour. Nothing had happened in the intervening year or so.

I wrote a lot about the idea of a policy costing unit here pre-Covid when the idea was being worked up by the government and The Treasury (there was a formal consultation process at one point). My most recent post on the issue was here. I ended that post with this thought

It won’t improve policymaking, it won’t change the character of elections, but it might –  at the margin –  create a few more jobs for economists.

I remain staunchly opposed. This was an extract from a submission to the Treasury consultation

Parties have adequate incentives already to make the case for their policies, in whatever level of detail the political (voter) market demands, and… already have access to the Parliamentary Library resources, parliamentary questions, and Official Information Act requests.  A policy costing office –  not found in any small OECD country –  would be, in effect, just a backdoor route to more state funding of parties (and not necessarily an efficient route – bulk funding would be preferable if state funded was to be more extensively adopted).  It also reflects a “inside the Beltway” conceit that specific costings are highly important, and that use of a single “model” or set of analysts somehow puts everyone on equal footing  (it doesn’t –  public service analysts having their own embedded assumptions about what is important, what behaviours are sensitive to what levers etc.)   With the possible exception of the Netherlands, I’m not aware of any country where a political costings office products plays any material or sustained role in election campaigns and outcomes.

Here was the list of other reasons from that 2019 post

I’ve listed most of my objections previously, but just quickly:

• there isn’t an obvious gap in the market.   At present, political parties produce costings (sometimes reviewed by independent experts) to the extent they judge it to be in their own interests to do so.  Voters, in turn, can judge whether the presence or absence of any costings, or any debate around them, matters much.  Existing parliamentary parties have access to considerable taxpayer resources which they can draw on to develop and test policy proposals,
• it isn’t obvious when, if ever, a New Zealand election in at least the last fifty years has turned on the presence, absence, quality (or otherwise) of election costings.  It is a technocratic conceit to suppose otherwise: people vote for parties for all sorts of reasons (values, mood affiliation, fear/hope, being sick of the incumbent, trust (or otherwise)) which have little or nothing to do with specific policy costings,
• the relevance of specific policy costings (and indeed overall fiscal plans) is even less under MMP than it was in years gone by.  Party promises are now little more than opening bids, as coalitions of support are put together after the election to govern (and on almost every specific piece of legislation).  We simply aren’t in a world where a few dominant ministers dominate a Cabinet which in turn has a majority (or near so) in the government caucus, which in turn has an unchallenged majority in Parliament,
• the “fiscal hole” argument (from the 2017 campaign) remains an utter straw man in this context.   First, when Steven Joyce made his claims in 2017 lots of people, including experienced ex-Treasury officials, weighed in voluntarily, and debate ensued about whether, and in what sense, Joyce was saying something important.  The system –  open scrutiny and debate –  worked.  And, secondly, a policy costings unit –  of the sort the government apparently envisages – would not have made any useful contribution to such a debate, which was about the overall implications of Labour’s fiscal plans, not about the costs of specific proposals Labour was putting forward.     Elections are messy things –  always were and probably always will be, and that isn’t even necessarily a bad thing.
• some of the arguments made for a policy costings unit might have more traction if, somehow, every political party and candidate could be forced to use it (say, submit all campaign promises to the costings unit at least three months prior to an election, with the costings unit issuing a report on all of them say at least one month prior to an election).  But even if you thought that might be a good model, it isn’t going to happen (and there is no credible way that such a model could be enforced).  Instead, the proposed costings unit will be used when it suits parties, and not when it doesn’t, and will probably be most heavily used by parties that are (a) small, (b) cash-strapped, and (c) like to present themselves as policy-geeky.  The Greens, for example.  One might add that the unit would most likely be used by parties that believe their own mindset is most akin to that of those staffing the unit –  likely to be a bunch of active-government instinctively centre-left public servants.  Embedded assumptions can matter a lot –  The Treasury used to generate wildly over-optimistic revenue estimates for a capital gains tax, and it was probably no coincidence that as an agency they supported such a tax. 
• the policy costings unit seems, in effect, to largely represent more state-funding for (established) political parties.  That might appeal to some, but even if you thought more state funding was a good idea (and I don’t) it isn’t obvious why this particular form of delivery is likely to be the best or the most efficient.  Money might be better spent on research and policy development (say) rather than “scoring” at the end of the process, for detailed plans that will almost inevitable change before they are ever legislated.  And if we want to spend more on policy scrutiny, I reckon a (much) stronger case could be made for better-resourcing parliamentary select committees.
• the interim proposal for next year’s election would enable only parties already in Parliament to utilise the facility.  Again, this has the effect of further entrenching the advantage established parties have in our system (I hope it will be re-thought when the legislation itself is considered).
• practicalities matter: there probably won’t be much demand on a policy costings unit in the year after an election, and could be quite a bit in the year prior to one.  How then will be unit be staffed and a critical mass of expertise maintained?  If people are seconded in from government agencies, would we really have an independence (including of mindset and model) at all?  And costings skills aren’t readily substitutable with bigger-picture fiscal policy (or macro policy) analysis skills.
• the lack of transparency around the proposed institution should be deeply concerning.  As far as I’m aware there has not yet been any indication as to whether the policy costings unit would be subject to the OIA (as the Auditor-General and Ombudsman are not, and nor is Parliament more generally).   The Minister of Finance has indicated that any costings the unit did would only be released with the consent of the political party seeking the costings.  That should be a major red flag.  In my view, any new unit should be (a) explicitly under the OIA, and (b) the enabling legislation should require that any costings done for political parties should automatically be released 20 working days after being delivered to the relevant political party (or more quickly if the costing is delivered within 20 days of an election).  A policy costings unit should not be a research resource for political parties – the only possible basis for confidentiality – but a body that at the end of the process provides estimates based on the details the relevant party has submitted. (As I understand the system in Australia, costings provided during the immediate pre-election period are automatically released, but others are not.)

The fourth bullet there – re the 2017 “fiscal hole” debates – is germane to Willis’s claims in the last 24 hours. The sort of policy costings unit that has been proposed costs specific policy proposals, but does not provide reports on the coherence or otherwise of overall fiscal strategies. The presence of such a unit would have made no obvious difference to anything about the furore around the CTU report.

From a more narrowly political perspective, one might also note that if National is really championing this new source of employment opportunities for economists, and (which is what it is) additional state funding for political parties, it isn’t a great signal of the seriousness of their commitment to fiscal restraint. Renney might, after all, have the beginnings of a point.

Incidentally, in this rather silly political debate, Grant Robertson emerges no better than anyone else. He is reported as having said that National’s change of heart on a policy costings unit had meant it was too late to have done anything for the 2023 election. Except that he himself in 2019, announcing the new policy costings unit policy Cabinet had just agreed to about a year out from the 2020 election, explicitly announced a transitional non-statutory arrangement for the 2020 election, pending full establishment in 2021. If Willis’s change of heart was a year or more back, presumably the same could have been done this year. (I’m glad he didn’t of course.)

Finally, I have seen this morning at least one commentary suggesting that independent fiscal institutions are now the way of the world, the OECD champions them etc. A policy costing unit is not really what most countries – or international agencies – have in mind in advocating such institutions, which are typically more about independent monitoring and reporting on government fiscal strategy and policy (ie macro in focus). Very few countries have state-funded policy costings units, none of them small and (by advanced country standards) relatively poor.