Scenario analysis: the boulder no one has moved

Ask the people preparing Australia's first mandatory climate reports what worries them most, and one answer tends to come up before any other: scenario analysis. It is required, it is hard, and it is the part of the regime practitioners are least settled on. It is worth being honest about why, because pretending it is solved helps no one.

Under AASB S2, an asset owner has to assess the resilience of its strategy using climate scenario analysis across at least two pathways, one around 1.5 degrees and one well above 2. The intent is sound: test the portfolio against different climate futures and show you understand what each would do to it. The trouble is in the doing.

Start with the scenarios themselves. The reference futures most often reached for, the Shared Socioeconomic Pathways and the NGFS pathways, were built to model the climate and the macro-economy, not to value a diversified investment portfolio. Few people believe any single one of them will come to pass. Used as prescribed – often as low-probability bookends – they ask a fund to plan around outcomes it does not consider the most likely, which is the opposite of how an investor reasons about risk.

The deeper mismatch is structural. Scenario analysis as the financial world built it works reasonably well for a single business – a steel maker or an energy company – where you can trace one pathway through one set of assets. An asset owner holds a slice of the whole economy. A credible scenario for steel is not a credible scenario for transport or technology, and each sector's path turns on policy across a dozen jurisdictions. Build it properly, sector by sector and geography by geography, and the aggregation back up to a single portfolio number becomes so assumption-laden that it stops meaning much.

Then there is consistency. Run the same portfolio through the same scenarios using two different providers' models and the results can come out markedly different. For a figure that is meant to be decision-useful and comparable, that is a real problem. Faced with an output they do not trust, preparers do the rational thing: they publish it with a disclaimer telling readers not to rely on it, or they decline to quantify at all. That is the made-safe reflex we have written about across this series, and scenario analysis is where it bites hardest.

None of this is an argument against the requirement. It is an argument for being honest about what scenario analysis can be. It is not a forecast and treating it as one is the mistake. It is a blunt instrument for stress-testing exposure, and a blunt instrument that is transparent, consistent and comparable is worth far more than a precise-looking one that no two providers reproduce.

Emissions estimation went through the same passage. The industry stopped chasing false precision and accepted standardised, transparent estimates that everyone could compare, and the disclosure got more useful, not less. Scenario analysis needs the same move: accepted reference scenarios applied consistently, assumptions on the table, and a clear view of what the fund thinks is most likely alongside the bookends.

The more useful emphasis is already shifting from a single dollar figure toward resilience. Which holdings are adapting, which sit out of harm's way, which have credible transition plans. That is a question an asset owner can actually answer and act on, and it speaks to members more honestly than a contested number ever will.

Whatever approach a fund settles on, the output is only ever as good as the exposure data underneath it, and that is where the weakness is greatest. Physical risk is the most under-modelled part of the picture precisely because the geolocation and asset-level exposure data feeding the models is the thinnest. The scenario engine, and the house view that drives it, belongs to the fund. What has to be solid first is the layer beneath it: reconciled holdings, with emissions and physical exposure traced back to source. That is the part Pathzero works on, so that whatever scenario a fund runs, it is running it on data it can stand behind. The boulder will move. It moves faster once the ground underneath it is firm.