Tuesday 23 February 2021

Blessed are they that have not seen the model, and yet have believed

The Climate Change Commission's recommendations span the breadth of the economy. They are required to come up with sector-by-sector climate budgets consistent with getting New Zealand with net zero emissions under the Zero Carbon Act. 

The sector-by-sector budgets rest on underlying models. The models build predictions about what will happen as ETS prices rise, and what will happen when some additional constraints are put into the system. Some of the CCC's recommendations then mandate what they think are their best guesses about what a carbon price would do, subject to those constraints.

The scope is vast. The entire economy, really. 

And the Government has already signaled that it will just do whatever the Commission says to do. 

So getting things right seems to matter and is rather high stakes. 

In that kind of situation, you'd think that the underlying models would be available for checking and testing. Getting bits wrong could be really really expensive, whether you want to frame it as economic costs, or as carbon mitigation forgone. 

But the Commission is not in a sharing mood. Here's Kate MacNamara.

Critically, the commission has not provided either sensitivity analysis nor the marginal abatement costs, broken out by industry.

That data matters. Sensitivity work helps economists to understand just how precarious that "less than 1 per cent of GDP" figure is. Will it alter significantly with slight adjustments to inputs? And the industry data for abatement cost would allow interested parties to properly test the assumptions the commission has made.

A commission spokesperson said it was unable to answer questions before the Herald's deadline on Monday. But one reason Carr has given for withholding information is the use of some US$6000 worth of proprietary global trade data from the Department of Agriculture Economics at Purdue University. It isn't clear why this data can't simply be stripped out.

In an emailed response to the letter signatories, Carr also said the board of the commission will consider the request for more time. But he didn't sound hopeful.

"As you know the commission has a deadline to deliver its final advice on or before 31 May 2021. Evaluating submissions and determining the impact of submissions on our draft advice also needs adequate time," he wrote.

You might have hoped that plans that have potential to re-engineer the entire economy would have more provably robust underpinnings. 

A couple years ago, our shop started putting out some new measures on school performance, based on work in the SNZ datalab by our excellent Joel Hernandez. We put up a technical report to go with it to show our workings, and we made all of our code available to anyone else in the datalab to check over and build on if they wanted to. 

If Joel's work turns into policy, we'd be really happy about it, but there isn't a direct channel there.

The Government has signaled they'll just do whatever the Commission tells them to do. But the Commission has done less to make their workings available for testing and checking than we did for a report on school performance. Our remit was small. Theirs is, potentially, the re-engineering of the entire economy. 

Blessed are they that have not seen the model, and yet have believed the results as described by their analysts in a series of webinars. 

No comments:

Post a Comment