*Submitted on February 1, 2021*

We now know that the Tiwai smelter is likely to be around for another four years to December 2024; but prior to 14^{th }January we didn’t know this. What do futures prices tell us about futures traders’ views of the probability of Tiwai staying longer than August 2021?

*By Greg Sise, 1 ^{st} February 2021*

There is a big difference in spot price expectations between having Tiwai in the market versus closed, by virtue of it being such a large load: Tiwai in the market, prices high, Tiwai out, prices low. These price expectations have an impact on futures prices, because futures prices (and all electricity hedge contracts) settle against spot prices, and therefore their prices change as expectations of future spot prices change.

For example, the price of the quarterly futures contract for Benmore for the quarter expiring 31^{st} March 2021, reflects expectations of the average spot price at Benmore for that quarter. When those expectations change, the futures prices change. These spot price expectations are not perfect, and there is ample evidence that futures prices can sit higher or lower than one might expect based on a completely unbiased estimate of futures spot prices, for quite long periods. But even as these biases are maintained, futures prices will still move in response to changing expectations.

So now that we have more certainty about the future of Tiwai, I thought it would be interesting to look back and estimate how likely futures traders thought closure might be. It turns out that we can do this using the futures prices themselves, albeit with some assumptions that hopefully aren’t too crude.

The first assumption is that immediately before the announcement on 14^{th} January (that Tiwai would stay until end of 2024 instead of closing by the end of August this year), the futures prices in each quarter reflected a weighted sum of the price traders thought would apply with Tiwai, and without Tiwai:

Futures price = A x futures price with Tiwai + B x futures price without Tiwai.

In this little formula, A and B are just the probability of each of the two states of the world, one with Tiwai and one without.

The day after the announcement, we can also assume the futures price for each quarter was equal to 100% of the futures price with Tiwai to end of 2024.

We only need one more number in order to estimate A, the probability that traders gave to Tiwai staying: the futures price in each quarter with Tiwai definitely closing by end of August 2021.

But this price was not directly observable on the date of the latest announcement, 14^{th} January.

However, we can make a crude estimate of how much lower the Tiwai-gone price would be than the Tiwai-staying price, by looking at the fall in futures prices immediately after the original closure announcement on 9^{th} July.

Confusing? Yes, but bear with me.

With the assumptions above, and a little bit of math, we can make some rough estimates of the probability that futures traders gave to Tiwai remaining in the market in each quarter, and they are shown below, including a curve fitted to the quarterly plots. We have used the prices at Benmore, rather than Otahuhu, because the price differences with Tiwai closed would be larger there.

What the chart suggests, is that traders’ aggregate probability of Tiwai staying fell progressively from Q1 of this year (Mar-21 expiry) through to the last quarter that could be traded back in July, which was Dec-23, where the probability was as low as 20%.

Even with Tiwai closing it would have remained in the market to some extent through into the Sep-21 quarter, so you might think the probability would be constant from Mar-21 to Sep-21, but what we may be seeing is an expression of the difference in price between it being fully in the market versus partially in the market.

These probabilities can be compared with our own forecasts, in which we read the signals from government and Tiwai as indicating the smelter would either be in for at least two years, but less than about four years. So, our estimate of the probability of Tiwai staying was 80% through to the start of a progressive phase-down of demand from Tiwai, starting at various times in 2023 (varying by forecast scenario).

In other words, our probability of staying was fairly constant for the first three years, whereas the chart suggests traders’ views of probability of Tiwai staying allowed for it to be closed anytime from August 2021 to end of 2023.

So, other than of being of passing interest, what does this progressive fall mean in the wider context of the futures market?

What it suggests to me is two-fold: first, rather obviously, the futures traders collectively had a different view of the closure probability profile than we did.

Second, our assessment of the probability of Tiwai staying is just one forecast, whereas the futures traders presumably would have a range of forecasts between them. Isabelle might be thinking closure in 2021, Brent might be thinking end of 2022, Trev and Amy mid-2023, and so on and so on. So, if you assume that the futures market is not dominated by one or two large players, then the futures prices will inevitably come to reflect some sort of average probability in each quarter.

Putting it another way, the futures prices reflected a diverse range of views on how long Tiwai might actually remain in the market, which were all “smoothed out” in a way. Now that is interesting, because we felt the signals for staying at least two to three years were quite strong. But now it seems that not everyone thought that, far from it.