Friday, 28 August 2026

Hating trees

New Zealand's Zero Carbon Act sets net emission targets. 

This is sensible. The atmosphere does not care whether the next tonne of CO2-e is not-emitted or removed. They amount to the same thing. 

It also aligns with the ETS's design, which requires surrender of an NZU for putting CO2-e into the atmosphere and awards an NZU for pulling a tonne of CO2-e out of the atmosphere. 

Arguably NZ ought to have had a carbon tax in the first place and a subsidy for removals, but it's hard to see it's worth the switching costs. 

There are things I would improve in the ETS. 

I would legislate the quantum of *unbacked* NZU that can be auctioned or allocated between now and 2050. That quantity, plus any existing stockpiles of unbacked NZU still held in reserve, would represent net emissions from the covered sector from now until forever. 

I would replace the price cap with one that tracks global weighted average carbon prices in credible systems. If prices here ever hit that cap, the government could purchase units in the cheapest country in the bundle that defines the average, retire them, and issue an equivalent quantity of NZU. Those units are backed by reductions elsewhere. NZ carbon prices would never exceed international carbon prices. 

And I'd rejig industrial allocations to ditch the 'reduces by 1% per year no matter what' with 'scales down commensurately with emissions intensity in carbon-leakage-relevant markets'. And look closely at CBAM mechanisms to make up differences. I think one was likely warranted in cement.

But none of those would take forestry out of the ETS. Trees sequester carbon. It's removed from the atmosphere as trees grow. The ETS deems all carbon immediately returned to atmosphere on harvest, which is clearly wrong, but the averaging used for credit award might be a bit too front-loaded - I've not gotten deeply into the weeds on that. Maybe it could use some fine-tuning. And there's a defensible case for surrender-or-replant liability falling onto the land on which a carbon forest sits if the company awarded NZU while trees were growing has folded before the surrender-or-replant obligation hit. 

Those are all just tweaks. The bones of the thing are fine - other than the 'not having a durable intertemporal cap' part, which is eminently fixable. 

The system finds its own balance. When it's cheaper to harvest carbon from the atmosphere than to prevent its emission, people will plant trees and generate NZU. The more that that happens, the higher the NZU-price necessary to draw the next hectare from whatever other use into carbon forestry. The land with the lowest combined opportunity cost and conversion cost converts first, then the land with the next-lowest cost, and so on up the track. And tech for reducing gross emissions keeps improving. Hard to see combustion engines continuing for much longer at the pace of battery development, so long as trade with China can continue. 

Simon Upton wants to ban trees from the ETS and perhaps only allow them for offsetting agricultural emissions

Describing the 18 year-old emissions trading scheme (ETS) as “rudderless”, Upton released new modelling with insights on how excessive reliance on forestry plantings leads to a collapse in the scheme’s NZ Unit price on a tonne of carbon by the mid-2030s while achieving very little at present.

Ok. Recall that NZUs are durable and don't expire. And that people investing money to plant carbon forests aren't idiots. If a carbon forestry conversion only makes sense if you can sell NZU for a particular price, you won't invest in conversion if you expect the price to be below that. 

No carbon forester has to sell their NZU when they receive them. They can hold them. That means the carbon price path ties to the interest rate. If I expect that NZU will be worth 8% more next year than this year, and selling an NZU today to invest in a bond only gets me 4%, then I'll hold the NZU. Carbon prices today go up relative to future NZU prices. The price path should wind up roughly following the interest rate. 

And, if carbon prices *did* unexpectedly collapse, carbon foresters have another option. Suppose you planted a while back and have already harvested NZU that you've then sold. If the price of carbon credits drops substantially, you can fell your carbon forest, sell the timber, and buy cheap NZU to meet your surrender obligations. 

“The modelling shows that New Zealand’s net emissions would drop below zero around 2040, meeting the first leg of the 2050 domestic target of net zero early.

“However, net emissions are projected to only stay below zero until the mid-2050s. That means that the second leg of the domestic target – that net zero is maintained ‘for each subsequent calendar year’ would not be met,” the report, tabled in Parliament today, says.

Titled “Adrift: What future does the ETS have?”, the report makes numerous recommendations for reform, all with a focus on changing how forestry is treated.

“There will be a cry from the (forestry) sector to make it stable and certain but what we are seeing is that without change, it won’t work,” said Upton.

If people hold NZU and redeem them after 2050 rather than before 2050, that isn't any kind of problem. So long as the government is not issuing or allocating unbacked NZU after 2050, and NZU surrendered after 2050 are either ones generated through sequestration, or ones that someone held onto rather than redeeming before 2050. And it is strictly better for a net tonne of emissions to happen later rather than earlier. Carbon accumulates. An NZU surrendered today is more years of higher net emissions than an NZU surrendered in 2070. 

Upton identified permanent forests as particularly prone to abandonment and limited management since they stop producing ETS revenue when they are mature. However, they remain a major fire risk and have to be maintained in perpetuity if the carbon they store is not to be released.

If you have a registered carbon forest and it burns down, you have two options. You can surrender NZU to cover the carbon that was released to atmosphere, or you can replant. If you replant, you do not get credit for re-sequestering the carbon that your forest had just released. 

It's fine, unless the surrender obligation falls on no one, carbon prices are high, and replanting has become expensive. But that is also solvable. For new forest registrations, require that surrender obligations follow through to the land if all else fails. In that case, the land's owner picks up the obligation. 

It feels like a whole lotta folks really wish that we did not have legislation targeting net emissions or an ETS targeting net emissions. That they think gross emissions are the sin, and sequestration is some form of sale of indulgences. And that because it turns out that planting trees is a very cost-effective way of sequestering carbon, we're somehow enabling sin. 

But this isn't and shouldn't be a religious mission. It's a tech problem that eminently solvable. Target the ETS at its one big job of reducing net emissions in the covered sector, make sure that it's as strong as possible, and use other policies if you want to target other stuff. 

Monday, 24 August 2026

An economics without trade-offs

One reasonable definition of economics is the study of choice under conditions of scarcity, or choice under constraint.

When getting more of one good thing requires forgoing some other good thing, what do people do? How do they decide? 

Ganesh Ahirao (was Ganesh Nana; last name changed - I think because a historic error in NZ systems was finally more recently corrected) proposes an Economic Governance Act to replace core parts of the Public Finance Act.

Jack Tame asked how part of that would work. If the Minister of Finance were deciding how to spend the last billion dollars of available funding and were choosing between a hospital infrastructure project and a wetland restoration project, how would the Act guide that choice? 

Both projects would be supported under various parts of his proposed principles of responsible economic governance. The hospital would count under both the 'being a good ancestor' provision's 1(a)(ii), physical infrastructure and facilities, and under the 'social floor' provision's 1(b)(iii) access to health. The wetland project would count under 1(a)(i)'s 'healthy natural environment and associated eco-systems'.

There are seven objectives across the two domains, many of them with multiple parts. And there's no ranking. So how would the Minister decide? If the Act provides no guidance, what's the difference from the status quo? If it does have a concrete way of dealing with these trade-offs, where is it?

Ganesh didn't like the question, noting "What you're saying is the country can't afford both." 

The question very specifically asked about the last billion in public funding. 

Tame went on to ask what stops the next Muldoon if the fiscal responsibility provisions are stripped from the Act; Ganesh said that's 40 to 50 years ago and now irrelevant. 

And, somehow, the credit agencies would reward NZ for tearing up the fiscal responsibility provisions that have staved off a downgrade - so long as the government could make the case to the credit ratings agencies. 

A world without trade-offs. And a world where, somehow, money and borrowing costs never matter. It starts edging toward MMT.

A couple of other things seemed a bit puzzling in his framework.

Every environmental regulatory regime has to balance environmental harms against economic costs. Both are purposes in his proposed Bill. Which section tells the regulator how much weight to put on 2(2)’s productive and prosperous economy as compared to 2(3)’s kaitiaki role? And similarly for 4(1)(a)’s healthy natural environment against social licence in places that rely on that economic activity – or both of those against aspects of 4(1)(b)’s social floor that are funded through taxes raised on those activities? How does his system deal with trade-offs other than by denying they exist?

His Act provides no limitation clause comparable to the RSB's clause saying it creates no claims at law. Without that clause, could someone without a home sue the government for failing to provide one (part of the social floor) or for preventing his building one on his own land with his own money (zoning)? If suit is possible on the former, how could Treasury account the potential liability, or have we just given up on accounting?

The defensible version of an emphasis on broader outcomes reduces to Bill English's social investment approach. Run a broad CBA across long-term effects of current initiatives, and use that yardstick comprehensively across spending areas. That version takes trade-offs seriously. 



Around the traps

A few bits I've neglected to blog.

Friday, 14 August 2026

Reader mailbag - prediction markets

In today's inbox:

Hello Eric,

My name is [redacted].  I read your article about prediction market regulation with interest.

As a student in 2013 I placed a single $60 bet which brought down iPredict from a run on the market related to Peter Dunne, which I intentionally caused based upon groupthink and illogical beliefs in the truthiness of the insider betting ring. I knew exactly what would happen when I joined and placed the well timed bet. I knew what I would prove, and let the energies of my opponents be redirected into their own damage.

iPredict is long gone, a decade ago, RIP. It was fun, but I guess if it had lived longer it could have competed with Kalshi and Polymarket. But now I live in the United States, and there are wildfires, and people are betting on wildfires on "prediction markets". How can you endorse this.

What is big may fall, what seems consensus may be false, and ultimately: I will be watching. I look forward to your next article about moral hazard.

My reply:
iPredict was always fun like that. Folks would convince themselves that a spike was due to an insider, and sometimes it was, but sometimes it wasn’t – whether a noob trader who placed a dollar-value order without checking the book, or someone just having a lark at low dollar stakes. But the markets proved remarkably accurate overall: trades at $0.75 turned into contracts paying out at $1 about 75% of the time. 

It’s been amazing to see what Kalshi’s been able to build in a world without deposit limits. 

The main concern I’d have on wildfire markets would be whether they’d encourage a very bad kind of insider trading. I don’t think Kalshi has any wildfire markets; their natural disaster markets are all on completely exogenous events. Polymarket has had those; best I’m aware, they’re not yet CFTC-authorised. 

I don’t know how material the risk is. I mean, a slightly less direct route would be to short insurers with exposure to that risk before starting fires. Similarly for a lot of the other ‘it will encourage them to do the bad thing’ risks: there are generally already very thick financial markets where options trading could get you similar results. I don’t think Trump needs prediction markets to cash in on Trump-induced oil price volatility. Brent crude futures are enough. 

At the same time a large punt on oil futures can have many causes, including “I will need a lot of oil in a few months”. A large and suspiciously-timed punt on a prediction market can lead to questions of who made the trade, identification of the trader (you have to do your KYC to trade at Kalshi), and then inquiries. 

I really wish Kalshi would set a market on “giant Wellington earthquake”. I could pay a friend in the US to take a position for me and treat it as insurance on otherwise uninsurable local earthquake risks. I’ve long wanted parametric insurance on a Wellington earthquake, and that is mathematically identical to a prediction market contract on it. Maybe someday!
  • How many major Atlantic hurricanes will there be this year?
  • How many Atlantic hurricanes will there be this year?
  • Number of tropical storms in the Atlantic this year
  • How strong of an earthquake will occur worldwide before Sep 1, 2026?
  • Will there be an 8 magnitude earthquake in California before 2027?
  • 8.0 magnitude earthquake in Japan before 2030 [39%!]
  • Number of tornadoes this month
  • Major volcano eruption this year?
The California earthquake market has a 5% chance of the event, and just under $400,000 in volume. But the order book is still thin at reasonable prices. You could spend $400 and buy every contract in the book up to a $0.10 price, and get a $5100 payout if the event happens. Perhaps putting a giant buy order into the book would draw out liquidity. 

Thursday, 13 August 2026

Spills

A new and more rational basis for ongoing choice of party leaders in a Parliamentary system.

Step one. Set a prediction market contract. "Pays $1 if the leader of the X Party is a Minister after the election." 

Step two. Set prediction market contracts: "Pays $1 if the leader of the X Party as at the date of election is {Name}."

Step three. Set conditional contracts. "Pays $1 if the leader of the X Party is a Minister after the election conditional on {Name} being Party Leader as at the date of election."

Step four. Set a decision rule. "The leader of the X Party changes to {Name} if {Name} shows a demonstrable sustained improvement in the Party Leader being a Minister after the election over the status quo, and superior to other {Name} options."

Works for both major and minor parties; their leader only becomes Minister (or Prime Minister) conditional on being in a winning coalition.

Punters from all parties could weigh in and put their money-votes on who they think would be most likely to improve each party's chances. 

A crazy ideological campaign to try to tank the other side's chances by picking a bad leader would draw in liquidity from outcomes-based traders - that's Hanson & Oprea (2009). The 'demonstrable sustained improvement' part is to give time for liquidity to come in in response to attacks - as well as guard against blips. 

No more dramas. No more press conferences. No need to try to attempt to count to whatever the required number of MPs might be to effect a spill and trust that nobody changes their mind along the way. Just watch the prices and let the leader be whoever maximises the Party's chances of being in a winning coalition. 

Probably only feasible in a country that's sensible enough to legalise prediction markets. 

Monday, 27 July 2026

Geloso on how to think like a good economist

I really like Vincent Geloso's slide deck on market processes, market failures, and government failures.

Over the past couple of months, Claude read through about four million words I've written to develop an "Eric skill". I can now give it documents that I don't have time to read, and it'll tell me what to watch for. 

When I gave it Geloso's slide deck, not because I didn't have time to read it but because I really liked it, the things it thought were inconsistent with the Eric Skill were errors on its part, and helped further refine the Eric skill to avoid those errors. Basically - look more carefully back through the corpus for the bits it thought were inconsistencies, and update the skill accordingly. 

Anyway, the slide deck's here.

The main updates to the Eric skill consequent to its reading Geloso's slides:

What changed: four additions, all sourced from Eric's own corpus, not from Geloso's slides.

  • §1, endogenous excludability (line 15) — the one drawn most directly from existing material: his public-goods lectures (ECON 336, 224, 653; the public-economics notes) already state "excludability is a function of technology," with the scrambled-TV, congestion-charging, radio-tied-good, and Buchanan-club-goods examples verbatim. This closes a real gap where the skill under-represented his own teaching. Highest confidence.
  • §1, statogenic failure (line 29) — generalises his gas-ban and RMA/housing positions into a standing diagnostic. High confidence on substance.
  • §1, price-theory-and-discovery-as-one-investigation (line 13) — folded into the price-theory substrate as complementary layers, not a rival Austrian register, per Eric's explicit instruction this session.
  • §4, item 18, interventionist ratchet — the one most worth the red pen; it's the least directly attested in his existing wording, so it's my synthesis of his second-order-dynamics habit rather than a phrasing lifted from the corpus

Age of wonders.  

Friday, 24 July 2026

Let them build: electricity and datacentres edition

New Zealand's electricity companies know how to stick a pipe into the ground in the Taupo Volcanic Zone and generate electricity. They've been doing it for decades, and there is enormous untapped potential. 

America's hyperscalers are currently willing to pay a large premium for immediacy. They are sticking expensive off-grid generators beside datacentres that would have to wait years for a connection. 

If NZ could be the place where decisions on whether a power company is allowed to stick a pipe into the ground are made in weeks/months rather than years, and a similarly fast decision on whether a datacentre is allowed, tens to hundreds of billions of dollars could drop here in a very big hurry. 

My column at Newsroom this week suggested NZ should consider being that place. And that the window of opportunity will not be open forever. 

[As always, I didn't pick the headline]