Climate scenarios: carbon price shock sees asset prices slump

Crowdsourced scenario analysis suggests very few sectors safe from a post-COP carbon price pop

  • A sudden, dramatic increase in the cost of emitting a tonne of carbon would cause widespread dislocations across financial markets, according to Risk.net’s most recent crowdsourced scenario generation exercise.
  • Commodities and global equities would take a battering in scenarios where the average price of pollution permits in traded markets leaps by year-end.
  • Risk.net readers offered projections for a series of financial indicators across three price regimes: little or no change, a modest rise, and a spike.
  • But audience projections across all three regimes yielded wide distributions, indicating significant uncertainty among finance professionals.
  • The survey’s projections are seen as extreme, but not unlikely: a volatile and disorderly carbon transition is emerging as the most likely pathway, according to Zurich’s John Scott.

This article forms part of Risk.net’s series of crowdsourced scenario-generation exercises. Click here to download a PDF of the full results.

Delegates at this month’s UN climate conference have faced many difficult challenges. Some more unusual than others. A strike by Glasgow binmen has led to rubbish piling high in the streets, with giant rats reportedly running amok, terrifying diplomats and hardened locals alike.

In the apocalyptic future some imagine the summit’s failure portending, that could be the least of humanity’s worries. The Conference of the Parties to the United Nations Framework Convention on Climate Change – COP26 for short – is seen as perhaps the last chance for nations to take co-ordinated, meaningful action to avoid a rise in global temperatures above 1.5°C. To a growing number of delegates, that means factoring in the cost of carbon pollution to all economic activity, globally.

The results

During October, Risk.net asked roughly 50 of its readers what they thought would happen to a series of leading financial indicators during the final quarter of 2021, to December 31, under three different carbon price regimes, using IHS Markit’s Global Carbon Index as a benchmark: little or no increase in the price of emitting a tonne of carbon, with the price remaining under $60; a moderate rise, up to $120; and a sharp increase, above $120. The scenarios (see link to PDF below), built by Sapiat, were modelled from these anonymised forecasts.

The indicators are: the S&P 500; the Euro Stoxx 50; the MSCI Emerging Market index; Brent Crude Oil; and the US Dollar Index, which measures the greenback’s value against a global trade-weighted basket of currencies.

Those predictions were aggregated and used to feed Sapiat’s model, with the impacts of the regimes simulated on an illustrative portfolio containing a wide range of asset classes including equities, bonds and commodities, as well as smaller allocations to alternatives. US equities make up 20% of the portfolio; Europe ex-UK equities, 7%; US government bonds, 10%; and global ex-US government bonds, 10%. The remaining asset classes, including regional equities and corporate bonds, real estate, hedge fund allocations and commodities and infrastructure funds, have weightings of between 3% and 5%.

Out of the total respondents: 47% thought carbon prices would remain under $60 at year-end; 43% thought they would be between $60 and $120; while 10% said prices would leap above $120.

Click here to download a PDF of the full results.

A downloadable spreadsheet containing the table and chart data is available here.

A sudden spike in the price of carbon – by one of several means (see below) – would constitute a supply-side shock, which analysts warn could make oil crises of the past pale in comparison. But perhaps the far bigger unknown is how markets would react: as Charles Donovan, visiting professor of finance at the University of Washington, puts it simply: “Nobody really knows what the effect of a very rapid rise in a global carbon price would be” – because it hasn't happened before.

In Risk.net’s latest crowdsourced scenario generation exercise, we asked 50 of our readers that very question. The verdict: a sudden, significant jump in the price of carbon would cause turmoil in financial markets. A hypothetical portfolio holding a broad range of assets (see table A) would be likely to plunge in value, with a mean expectation of an annualised 14.4% decline.

Things look far worse at the tail, however: taking the three-month expected shortfall measure for the portfolio, investors could face a collapse in investment values of more than a quarter, with the biggest contributor to the drop being a fall of nearly 10% in US equities. European and Asia-Pacific equities, particularly Australia, also see falls, and even private equity assets see modest declines. Only government bonds provide some modest cushioning, indicating a likely flight to quality, as observed following the Covid shock last year.

Risk.net readers were asked to offer estimates on the movements of a series of financial indicators under three carbon price regimes. In the first regime, the price of carbon remains somewhere below $60/tCO2 by the end of 2021, according to IHS Markit’s carbon index, which tracks a trade-weighted average of European and US carbon emission allowances. In the second, the average price lands somewhere between $60–$120/tCO2; in the third regime, a higher price of above $120/tCO2.

 

 

The short, sharp nature of the implied shock, Donovan adds, means it is unlikely investors and other market participants would be able to avoid absorbing some losses in those scenarios. But firms interested in examining the potential impacts of a price adjustment should consider additional factors beside the speed at which the price could change, he says. 

Perhaps more striking than the worst-case falls is the sheer spread of opinion between participants: figure 1 showcases the range of expected portfolio impacts for each of the regimes. Forecasts for regime one show a huge range, between a near-25% decline and the lower bound, and a 6% return, for optimists in the survey.

 

 

This wide distribution of returns is to be expected, says Donovan – who previously led Imperial College Business School’s Centre for Climate Finance and Investment – the spread is indicative of a good deal of justified uncertainty among respondents. 

“The width of that distribution is no surprise, because nobody knows what the effect of a very rapid rise in a global carbon price would be,” Donovan says. “It’s indicative of a lot of guesses, some of which are going to be really well-formed, and some [of which are] just throwing darts.” 

While the >$120 regime sees the steepest drops in the dummy portfolio’s assets, the more moderate $60–$120 regime sees its own unappetising declines; cash holdings take a beating, with only China equities making a positive return, and world commodities sink to -0.19%. In such a scenario, Donovan says, you might expect to see investors making cautious commodity bets and looking for ways out of sub-classes considered to be less substitutable.

“You have lots of potential dispersion within each sector,” he says. “You could think about steel and lumber being fungible in some construction settings; coal and fossil gas as a long-short; there are a number of long-short pairings. This is probably why US equities are highly negative [in the results],” Donovan adds. “Investors are able to understand when an entire economy or an entire benchmark is heavily weighted towards carbon-intensive activities. At a macro level, you sell the entire benchmark.”

 

 

Even the more moderate rise in carbon prices seen under regime two might cause alarm for holders of certain assets. In sectoral analysis provided by Sapiat, a number of interesting distributions are visible – the most dramatic tails belong to the >$120 scenario, but the $60–$120 regime isn’t without its share of significant movements. Hits to equity in various sectors could result in the middle scenario, the exercise suggests, with the most pronounced dip seen – unsurprisingly – in oil and gas, which could witness contribution to portfolio returns as low as -19%. Industrial equity could also fall sharply, some respondents said, as could financials and tech. 

Conversely however, some predict modest gains could await investors in oil and gas equity, the analysis suggests. Peak projected contribution to returns for the sector in both the $60–$120 and >$120 scenario stay above 10%, suggesting that some respondents, at least, have high confidence in the performance of carbon-intensive equity even in a world where carbon is expensive.

Such a range of views could indicate turmoil to come across the commodities complex, says John Scott, head of sustainability risk at Zurich Insurance Group.

“I think what we should expect to see – if we get the policies that drive an increase in carbon prices, and substitution and change in all these different sectors and services – is quite a lot of energy price volatility,” he says. “As time goes on, it becomes clearer that it’s unlikely that we’re going to have a managed, low-volatility transition, and I think this is what your survey is revealing: we’re not going to get a global transition in all these sectors in anything other than a disorderly fashion. It’s a competitive world, different markets are doing different things and it’s very unlikely that everyone’s going to do it all together.” 

The volatility seen in the third regime is a substantial 15.66, compared with 7.03 in the $60–$120 regime and 7.15 in the sub-$60 regime. In the highly volatile third regime, the healthcare and utilities sectors, while giving negative returns, perform the least badly.

Bonds fair much better across all scenarios than equities, notes Tim Wilding, Sapiat's head of research. "Interestingly, the volatility of the moderate rise scenario is lower than the other two, and the mean higher than the <$60 Carbon price scenario, suggesting strong confidence in the improved performance of bonds in this scenario."

Sapiat’s analysis also reveals interesting correlations between some asset classes. The China Equities class, for example, exhibits only a very limited relationship with the other assets held in the portfolio under regimes one – $60 – and two, $60–$120. A very mild positive correlation is found between it and the other equity types – spanning Australia, Canada, Europe ex-UK, Japan, UK, US and advanced emerging – as well as world private, equity, real estate, commodities, hedge funds and infrastructure, across all three scenarios.

 

 

China Equities are negatively correlated with some bond classes, but these correlations are still slight. A very small negative correlation is found between it, US government bonds and world ex-US government bonds under all regimes; in the third scenario alone, a positive correlation emerges between it and the corporate bonds types. The most significant positive correlations are found in the third and most extreme regime, where carbon prices go above $120; there, China equities has a 0.38 positive correlation with advanced emerging equities, and 0.35 with both Europe ex-UK equities, world private equity and world real estate.

Similar correlations can be seen to world commodities. The class is negatively correlated to three bond types – US government bonds, us corporate bonds and world ex-us government bonds – in most instances. In the >$120 regime, however, the correlation between world commodities and US corporate bonds turns slightly positive, reaching 0.06. As with China equities, world commodities’ positive correlations to the rest of the portfolio increase with the extremity of the regime. Its correlation to advanced emerging equities, for example, is 0.23 under regime one, 0.24 under regime two and 0.51 under regime three.

In Risk.net’s last crowdsourced scenario exercise, focusing on inflation, China equities tended to perform well in cases where other asset classes began to struggle. In the highest inflation regime, for instance, the class exhibited its strongest returns, while the rest of the portfolio – except for world commodities – slumped into the negative.

 

 

So just how likely are carbon pricing jumps like the ones seen in the scenarios? While calls for higher costs on carbon emissions have intensified in the run-up to COP26, influential voices in the financial community have been forecasting a rise for some time.

In June of last year, the Network for Greening the Financial System – a consortium of the world’s central banks – published scenarios indicating an average universal carbon price of $100/tCO2 would need to be set by 2030 for a decent chance of limiting global warming. And the Bank of England’s flagship climate stress test contains an “early action” scenario that sees the price reaching $900/tCO2 in the UK by 2050. In the “late action” scenario, it climbs above $1,000 by the same date. 

Few have predicted a price of more than $100/tCO2 by the end of 2021 – but real-life medium-term trends, however, suggest such a figure isn’t that far off: Ember’s price per tonne (EUR) was around €33 at the beginning of the year, and now sits at close to €59. If the price continues to increase at the same rate – that is, by roughly 78% around every 11 months – we could expect an approximate European carbon price of more than €100 by October 2022. 

“A lot would depend on what decisions are taken in the next two weeks at the COP,” says Zurich’s Scott.

He says it’s certainly possible that governments working in tandem could raise the price of carbon to the level in the most-extreme scenario: “If governments everywhere suddenly agree to create some economic driver – recognising that the IEA says you don’t have a material impact on economic decisions unless you’re in the range of $80 to $120 – and push tariffs or levies all together, I think there’s a chance.”

Still, as part of their COP 26 negotiating arsenal, European legislators have proposed an aggressive carbon border tax, which would slap tariffs on goods entering the EU from jurisdictions that don’t effectively tax carbon – which, according to the IMF, account for some 80% of global emissions. A uniform carbon price would apply uniformly for all emissions, globally. It could take several forms: a globally-agreed straightforward tax on carbon emissions; uniform tariffs on goods imported from regions that don’t impose pollution permits into those that do; or a floor on the price of carbon permits, which are already in place in some regions globally.

With a phase-in proposed from the end of next year, should the bill pass in its current ambitious form – or more likely, force governments from other nations to shore up their own emissions cap or tax regimes in response, for fear their exports become uncompetitive in Europe – the global cost of emitting a tonne of could spike dramatically.

Achieving this aim would be no mean feat: according to IMF estimates, some 80% of global emissions are unpriced.

And as Zurich's Scott points out, ‘possible’ doesn’t equate to ‘likely’: after nearly 30 years of COPs since the Kyoto Agreement, world leaders have yet to come to a decision that could yield such an impact.

 

 

 

 

 

 

The stress forecasts above show two sets of values, one per regime, calculated on a subset of responses to the survey, writes David Androsoni, chief executive of Sapiat.

In defining what a tail event is we need to look at the impact of those forecasts (eg, on a particular broad index). We picked the model allocation to capture a broad cross-asset class behaviour, and for comparability with previous analyses: broad global shocks are most relevant for climate scenarios.

Using each individual respondent’s set of answers for either carbon regime, we can measure the impact of that respondent’s scenario on the model allocation. Some responders may be forecasting that the mean of the model allocation is beyond its 95% value-at-risk, but the number of responses may be too few to be reliable.

We need to find a combination of respondents that satisfy these criteria of producing a robust, extreme but plausible and consistent stress scenario. We then need to run through all combinations, across all simulation samples, to find the cluster of respondent answers that satisfies these criteria.

Simply put, the criteria are that:

  • We can’t rely on one respondent: the results should be derived from a group response (ie, a subgroup of the crowd).
  • The responses within a group should be consistent with each other.
  • The views of each respondent and subgroup should reflect the most likely extreme, because we are exploring >95% (and the number of respondents in the sample is relatively small).

How the results were compiled

During October, Risk.net asked roughly 50 of its readers what they thought would happen to a series of leading financial indicators during the final quarter of 2021, under three different carbon price regimes: little or no increase in the price of emitting a tonne of carbon, with IHS Markit’s index remaining under $60; a moderate rise, up to $120; and a sharp increase, above $120. The scenarios, built by Sapiat, were modelled from these anonymised forecasts. Here, the firm gives a brief insight into its methodology.

Common to all crowdsourced scenarios is the assumption that forecasts from each individual respondent carry bias and uncertainty, but that their effects can be removed when using a large set of responses.

Sapiat applied the following settings during scenario construction:

  • Mean return. Where the mean return is being forecasted, each forecast is treated as independent, implying individual forecast errors are assumed to be diversified. This reduction of errors is in line with the so-called Wisdom of crowds (Galton, 1907).
  • Variance matrix. Since the respondents are assumed to have a good understanding of the context of inflation regimes and the impact on all inflation variables, we use the range of forecasts to estimate a variance matrix across respondents within each regime scenario.
  • Scenario combination. Finally, a combined estimate of the mean and the variance is set from the combination of all the scenarios. Each of the forecasts for a particular inflation scenario are weighted by the respondents’ probability estimate for that inflation scenario. These predictions are then adjusted using the forecast means and variance matrix from each inflation scenario before being weighted by the average probability of each scenario and aggregated together to get an overall forecast of future conditions.

The scenarios are then simulated over a single period forecast, ending December 31. Sapiat used the following proprietary engines in the simulation:

  • A regime model, which identifies probabilistically which regime the markets are currently following, and the likelihood of transitioning to any of the other regimes over the simulation horizon.
  • A simulation model, in which return paths are simulated by rigorously combining the forecast scenarios and the regime modelling over time. The resulting scenario distribution allows the calculation of scenario risk measures (including typical stress-test outputs for downside risk), but also plausible estimates for portfolio return.
    Where the forecasts include stress scenarios (defined as scenarios with large or unprecedented shocks), the resulting distributions include the simulated results of such shocks, and so the downside risk metrics for any portfolio may be calculated directly from the scenario distribution.
  • Interpretation. Sapiat employs expert judgement when turning audience views into future return scenarios. Since scenario distributions are simulated for all asset classes globally, and not just for those for which forecasts have been provided, the framework can be useful in modelling the returns of investment portfolios over multiple time horizons.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here