EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector

arXiv:2608.12363 · econ.GN, cs.AI, cs.CY, cs.LG, cs.MA, cs.SY, eess.SY, q-fin.EC · Submitted 2026-07-06 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: AI Radio. Generated commentary on the latest Artificial Intelligence papers.

Tom: Next we'll be talking about the paper "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector".

Jane: The paper was written by Javier Gonzalez-Ruiz, Carlos Rodriguez-Pardo, Alice Di Bella, Paolo Mastropietro, Jose Pablo Chavez-Avila et al. from Politecnico di Milano and CMCC Foundation - Euro-Mediterranean Center on Climate Change and RFF-CMCC European Institute on Economics and the Environment and Universidad Pontificia Comillas.

Tom: Stay tuned as we take you through the paper and discuss its implications.

Title: Tom: Welcome back to the show, everyone. Today we're digging into a paper that's been making waves in energy circles — it's called "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector." Jane, I have to say, that question mark in the title already tells you this is going to be a spicy one.

Jane: Oh absolutely, Tom. And it's timely too — this paper is basically responding to what Italy did in two thousand twenty-six with something called the Decreto Bollette. The government there decided to remove the carbon price from the bids that gas plants submit to the electricity market, so those plants look cheaper than they really are, which lowers the wholesale price consumers pay. Sounds great in the short term, right?

Tom: Right, and that's exactly the tension the authors are poking at. The paper comes from a big team — Gonzalez-Ruiz, Rodriguez-Pardo, Di Bella, Mastropietro, Chavez-Avila, and Tavoni — spread across Politecnico di Milano, the CMCC Foundation, and Comillas in Madrid. These are serious energy-systems people.

Jane: And they're asking a really pointed question: if you suppress that carbon price signal, what happens to the long-term incentives for building renewables and storage? Because the carbon price is what makes gas look expensive compared to wind and solar. If you hide that cost, you're basically telling investors that fossil fuels are cheaper than they actually are.

Tom: Exactly. And the paper's answer is pretty stark. They ran a detailed simulation of the Italian power system from two thousand twenty-five to two thousand forty and in most scenarios, suppressing the carbon price leads to significantly higher CO2 emissions — like, in some configurations, emissions go up by more than fifty percent compared to the baseline.

Jane: And the kicker is that the supposed cost savings for consumers mostly evaporate over time, because the ETS costs still have to be paid — they're just shifted into a tariff surcharge. So you get a short-term discount on your bill, but the system ends up dirtier and the long-term costs come back to bite you.

Tom: That's the trade-off in a nutshell. And the paper shows it's not just a theoretical concern — they modeled it with a really sophisticated multi-agent reinforcement learning framework called MARLEY, where generators and a policymaker all learn and adapt to each other's behavior over time.

Jane: Which is what makes this paper so compelling, Tom. It's not just a static calculation — it's a dynamic game where investors respond to the policy, the policymaker responds to the investors, and the whole system evolves. We'll get into how that works in a moment.

Tom: And we should — because the method is half the story here. But before we go deeper, let's just sit with that headline finding: a policy designed to lower electricity bills can end up undermining the very transition those same governments say they want. That's the tension we're going to unpack for the rest of the show.

Jane: Stay with us, folks — next we're going to look at what the paper actually found in its simulations, and why the results depend so heavily on what other market mechanisms are in place.

Summary: Tom: So we've set the stage — Italy's Decreto Bollette suppresses the carbon price signal for gas plants, and the paper "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector" asks what that does to the whole system. Jane, what did the simulations actually show?

Jane: Well, Tom, the headline is that the results split into two very different stories depending on what else is going on in the market. The paper tested five market configurations — from a bare-bones setup with just a capacity market, all the way up to what they call "hybrid" designs with aggressive contracts for differences for renewables and dedicated support for storage.

Tom: And in the bare-bones cases, the carbon price suppression is pretty devastating. Emissions jump by something like fifty-seven percent over the whole horizon in the capacity-market-only scenario. The system just keeps leaning on existing gas plants and builds more open-cycle turbines instead of solar and batteries.

Jane: Right. Because when you remove the carbon price, you're making gas look artificially cheap, and that erodes the revenue that solar and storage would have earned in the wholesale market. Investors see weaker returns, so they don't build. And the system gets locked into fossil dependence.

Tom: But here's where it gets interesting — in the most ambitious hybrid scenario, the one they call H-plus, emissions barely move at all when the carbon price is suppressed. The difference is tiny, like eight percent. So what's going on there?

Jane: That's the key insight. In that scenario, the government is running huge, sustained auctions for renewables and storage — we're talking contracts for differences covering up to one hundred twenty percent of average demand. When those mechanisms are big enough, they essentially replace the wholesale price signal as the main driver of investment. So even if you suppress the carbon price, the long-term contracts are still telling investors to build clean.

Tom: But that comes with a catch, doesn't it?

Jane: A big one. The paper is very clear that this only works if the government commits to that level of intervention permanently — year after year, for the entire fifteen-year horizon. And that's a fundamentally different market paradigm. You're basically saying the wholesale market doesn't really matter for investment anymore; the state is doing the planning through these auctions.

Tom: And that's politically awkward, because the same government that's willing to suppress the carbon price to lower bills is probably not the same government that's going to commit to massive, permanent green investment programs. The paper calls this out explicitly — the logic of the Decreto Bollette and the logic of the H-plus scenario are in direct contradiction.

Jane: Exactly. And there's another layer — they also tested a scenario where the suppression is temporary but uncertain. Investors don't know when the carbon price will come back. And that uncertainty alone is enough to delay investments and push up costs, even after the policy is reversed.

Tom: So the uncertainty itself is a tax on the transition. That's a really important finding for policymakers to hear.

Jane: It is. And it's one of those results that feels obvious once you see it, but nobody had actually modeled it this way before — with agents that learn and adapt to the policy environment in real time.

Tom: Alright, so we've got the results — now let's talk about what the paper suggests we actually do about it. That's coming up next.

Improvements: Tom: Welcome back. We're still on "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector." We've covered the findings — now Jane, what does this paper actually suggest we should do differently?

Jane: So the paper doesn't just diagnose the problem — it really pushes on the design of the long-term mechanisms. The authors argue that if you're going to have a hybrid market, with contracts for differences and flexibility auctions, those mechanisms need to be sized and sustained with real ambition. Not just a one-off auction here and there, but a durable, escalating commitment.

Tom: And they're careful to note that even their least ambitious hybrid scenario — the one they call H-minus — requires volumes that are comparable to Italy's actual two thousand twenty-five auctions. So this isn't some pie-in-the-sky fantasy. It's a scale of intervention that's already happening. The difference is that in their model, it has to keep happening every single year.

Jane: Right. And that's the improvement they're really advocating for — not just more mechanisms, but a credible, long-term commitment to them. Because their results show that the shielding effect only works if investors can reliably anticipate that the support will be there. If there's uncertainty about whether the auctions will continue, the investment signal weakens, and you get the same kind of damage as the carbon price suppression itself.

Tom: So the fix isn't just "spend more money" — it's "make a believable promise." That's actually a pretty profound point about how markets work.

Jane: It is. And there's another improvement they highlight, which is about the carbon price itself. They tested different carbon price trajectories — a flat seventy euros per ton, the baseline that rises to one hundred fifty and a steeper one that hits two hundred thirty by two thousand forty. And in the baseline without the Decree, a higher carbon price means lower emissions with only a modest cost increase. That's the standard story.

Tom: But with the Decree in place, that relationship inverts, doesn't it?

Jane: Completely. When the carbon price is suppressed, raising the carbon price just means higher ETS recovery costs passed through to consumers — without the investment response that would actually reduce emissions. So you end up paying more and getting nothing for it. The paper shows that in the suppressed scenarios, the system relies more on existing fossil assets, so the higher carbon price just becomes a tax on those assets with no clean alternative coming online.

Tom: That's a really stark result. It basically says the carbon price only works if it's allowed to work — if you suppress it, you can't just dial it up later and expect the same benefits.

Jane: Exactly. And the paper also flags some limitations that point to future work. They don't model cross-border electricity trade, which is a big deal for Italy since it imports a lot. And they don't model the demand side — no demand response, no electrification feedback. So the real-world effects could be even larger than what they show.

Tom: So the improvements they're suggesting are really about coherence — making sure the carbon price and the long-term mechanisms are working together, not against each other.

Jane: That's the through-line. And it's a hard message for politicians to hear, because it means there's no free lunch. You can't lower bills today without either accepting more emissions tomorrow or committing to a much bigger role for the state in planning the energy system.

Tom: Alright, so let's wrap this up — what's the big takeaway for our listeners, and what does this mean for the broader conversation about energy policy?

Conclusion: Tom: So we've spent the show on "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector." Jane, if you had to sum up this paper in a couple of sentences for someone who just tuned in, what would you say?

Jane: I'd say this: suppressing the carbon price on gas plants gives you a short-term discount on electricity bills, but it costs you dearly in the long run — higher emissions, weaker investment in renewables, and ultimately higher total system costs once the ETS obligations come due. The only way to avoid that damage is to have massive, credible, sustained long-term support for green investment, and that's a fundamentally different political commitment than the one that produced the Decree in the first place.

Tom: And that tension — between short-term relief and long-term transition — is really the heart of the whole thing. The paper shows it's not just a theoretical tension; it's baked into the market dynamics. Investors respond to signals, and when you distort the signal, they respond in ways that lock in fossil dependence.

Jane: Right. And I think the most important contribution here is the method. Using multi-agent reinforcement learning, they've built a model where the policymaker and the generators all learn and adapt to each other. That's a huge step beyond the static models that most policy analysis relies on. It lets you see these feedback loops — like how uncertainty about the policy duration alone can delay investment and raise costs.

Tom: And that's the kind of insight that should make policymakers pause before they reach for the quick fix. Because the paper is really saying: if you're going to intervene in the market, you need to understand how the market will respond — not just next quarter, but a decade from now.

Jane: Exactly. And the authors are clear that this isn't just an Italian problem. Any country with a gas-dominated power system and an emissions trading scheme could face the same dynamics if they try to suppress the carbon price signal. Spain and Portugal did something similar with the Iberian exception in two thousand twenty-two and the paper's findings would apply there too.

Tom: So this is a general lesson about market design, wrapped up in a very specific Italian case study. And it's a lesson that's going to become more and more relevant as energy prices keep spiking and governments keep feeling the pressure to act.

Jane: Absolutely. And with that, we're going to say goodbye to this paper — "EU-ETS under attack? The impact of carbon price suppression on the decarbonization of the power sector" — and get ready to look at what's next on the arXiv feed.

Tom: Thanks for listening, everyone. We'll be back soon with another paper, another set of questions, and another conversation about how research is shaping the world we live in. See you then.

Jane: Take care, folks.

Javier Gonzalez-Ruiz, Carlos Rodriguez-Pardo, Alice Di Bella, Paolo Mastropietro, Jose Pablo Chavez-Avila, Massimo Tavoni

Politecnico di Milano · CMCC Foundation - Euro-Mediterranean Center on Climate Change · RFF-CMCC European Institute on Economics and the Environment · Universidad Pontificia Comillas

econ.GN, cs.AI, cs.CY, cs.LG, cs.MA, cs.SY, eess.SY, q-fin.EC

Submitted: 2026-07-06

Updated: 2026-08-14

Comments: 58 pages, 12 figures, 12 tables. Includes supplementary material

Code: https://github.com/ray-project/ray

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 65/100

The gist: European countries are debating policies to mitigate the increased energy costs caused by renewed geopolitical tensions, while pursuing decarbonization and electrification.

Terminology

Actor — 34× in this paper · explained in 3 episodes
Reinforcement Learning — 28× in this paper · explained in 4 episodes
Action Space — 7× in this paper · explained in 1 episode
Deep Reinforcement Learning — 6× in this paper · explained in 2 episodes
Core — 5× in this paper · explained in 2 episodes
Non-stationarity — 4× in this paper · explained in 2 episodes
Robustness — 4× in this paper · explained in 5 episodes
Contract for Difference — 3× in this paper
Multi-Agent Proximal Policy Optimization — 3× in this paper
Action Masking — 2× in this paper · explained in 1 episode
Aggregation — 2× in this paper · explained in 1 episode
Benchmark — 2× in this paper · explained in 2 episodes
Expected Unserved Energy — 2× in this paper
Objective Function — 2× in this paper · explained in 2 episodes
Abstraction — 1× in this paper · explained in 1 episode
Capacity Remuneration Mechanisms — 1× in this paper
Consistency — 1× in this paper · explained in 3 episodes
Cooperation — 1× in this paper · explained in 2 episodes
Credit assignment — 1× in this paper · explained in 3 episodes
Decoupling — 1× in this paper · explained in 2 episodes
Energy Efficiency — 1× in this paper · explained in 3 episodes
Generative AI — 1× in this paper · explained in 2 episodes
Independent Proximal Policy Optimization — 1× in this paper
Loss Function — 1× in this paper · explained in 2 episodes
Multi-Agent Framework — 1× in this paper · explained in 3 episodes
Multi-objective Optimization — 1× in this paper · explained in 4 episodes
Nash Equilibrium — 1× in this paper · explained in 2 episodes
Partially Observable Stochastic Game — 1× in this paper
Regularization — 1× in this paper · explained in 3 episodes
Reward Function — 1× in this paper · explained in 2 episodes

Summary

European countries are debating policies to mitigate the increased energy costs caused by renewed geopolitical tensions, while pursuing decarbonization and electrification. A notable example is Italy’s 2026 Decreto Bollette package, which proposes to remove the carbon price equivalent from the bids of certain gas-driven power plants to wholesale electricity markets, among other provisions. We use this as a case study to assess the long-term implications of suppressing the carbon price signal in the electricity market for investment, emissions, and consumer costs. We employ a stylized Italian power system using MARLEY, a multi-agent reinforcement learning framework focused on long-term electricity market assessments. In this framework, we test this policy across configurations with varying levels of support for green investment, resource adequacy, and flexibility. Results show that partial suppression of the carbon price signal yields short-term cost reductions but only a minor long-term effect on total system costs, as the deferred emissions are ultimately repaid by consumers. CO2 Emissions rise across most configurations since suppressing the price signal erodes incentives for renewable and storage investment. Only the most ambitious configurations for supporting green investment avoid this outcome, but they do so by marginalizing the wholesale price signal itself, thereby requiring a commitment to a hybrid market paradigm that is in contradiction with the rationale of the proposed price intervention.

Improvements for AI systems

Based on the scientific paper, here are specific improvements I can make to AI systems and what the improved AI system can do:

Improvement: Implement the Independent Proximal Policy Optimization (IPPO) algorithm with the exact hyperparameter configuration from the paper (batch size 10800, mini-batch 1800, clipping 0.1, learning rate 5e-5, discount factor 0.995, GAE lambda 0.2)
What it can do: Simulate long-term electricity market dynamics (2025-2040) with 16 autonomous agents (8 incumbents, 8 entrants) plus a policymaker, capturing strategic interactions in concentrated wholesale markets
Improvement: Add a module that models partial carbon price signal suppression (as in Italy's Decreto Bollette) where ETS costs are recovered through end-user tariffs rather than embedded in generator bids
What it can do: Quantify the trade-off between short-term consumer cost reductions and long-term emissions increases, with the model predicting a 56.92% increase in CO2 emissions in the CRM configuration while only reducing system costs by 11.94% over the full horizon
Improvement: Incorporate the paper's [UD] scenario logic, where agents are trained across four possible policy reversal timings (years 4, 8, 12, or never) to prevent anticipation of exact restoration
What it can do: Model how regulatory uncertainty affects investment delays, showing that costs rise to levels equal to or above baseline once the carbon signal is restored, with delayed green investments that only partially recover
Improvement: Implement the full hybrid market framework with four investment channels: merchant, CfD auctions, capacity market (Reliability Option design), and flexibility auctions, each with distinct financial settlement mechanisms
What it can do: Test how different levels of green investment support (60%, 80%, 120% of demand targets) shield the system from carbon price suppression effects, with the [H+] configuration showing almost no emissions increase
Improvement: Add the three-state Markov regime-switching model for natural gas prices (Normal, Shock, Recovery) with log-normal peak multipliers (median 2x, range 1.3-3.5x) and expected crisis duration of 12 bimesters
What it can do: Predict system vulnerability to gas price shocks, showing that carbon price suppression increases vulnerability by 2-3x in most configurations, and quantify the cost impact of such shocks on different market designs
Improvement: Integrate the paper's validation methodology: Welch's t-test, Mann-Whitney U test, bootstrap confidence intervals (10 4 resamples), Holm correction for multiple testing, and Hedges' g effect size classification
What it can do: Automatically distinguish statistically significant but practically negligible differences from large effects, preventing false policy conclusions from simulation noise
Improvement: Add the capability to test multiple carbon price trajectories (flat 70 e/tCO2, linear to 150 e/tCO2, steep to 230 e/tCO2) and identify when the standard carbon-pricing trade-off inverts
What it can do: Show that under carbon price suppression, higher carbon prices increase system costs without reducing emissions (inverted relationship), as ETS recovery costs fall on a fossil-dominated mix

Predict that suppressing carbon price signals yields short-term cost reductions of 15% but leads to 35-99% emissions increases over 15 years, with the trade-off worsening over time
Identify which market configurations (CRM, H-, H, H+) can withstand carbon price suppression, showing that only the most ambitious hybrid designs (120% RES target) avoid emissions increases
Quantify the political economy tension: systems with strong green investment support (H+) are unlikely to also implement carbon price suppression, as the policy logic contradicts itself
Model the dynamic response of policymakers who endogenously adjust auction volumes in response to merchant investment declines, with CfD volumes rising to compensate when carbon signals are suppressed
Evaluate system resilience to gas price shocks across different policy and market configurations, showing that suppression increases fossil dependence and shock vulnerability
Provide statistically validated policy recommendations with effect sizes and confidence intervals, distinguishing robust findings from simulation artifacts
Simulate the full feedback loop between carbon pricing, investment incentives, emissions, and consumer costs, capturing the self-reinforcing dynamics that lead to fossil-fuel lock-in when signals are distorted

Abstract

European countries are debating policies to mitigate the increased energy costs caused by renewed geopolitical tensions, while pursuing decarbonization and electrification. A notable example is Italy's 2026 Decreto Bollette package, which proposes to remove the carbon price equivalent from the bids of certain gas-driven power plants to wholesale electricity markets, among other provisions. We use this as a case study to assess the long-term implications of suppressing the carbon price signal in the electricity market for investment, emissions, and consumer costs. We employ a stylized Italian power system using MARLEY, a multi-agent reinforcement learning framework focused on long-term electricity market assessments. In this framework, we test this policy across configurations with varying levels of support for green investment, resource adequacy, and flexibility. Results show that partial suppression of the carbon price signal yields short-term cost reductions but only a minor long-term effect on total system costs, as the deferred emissions are ultimately repaid by consumers. CO 2 emissions rise across most configurations since suppressing the price signal erodes incentives for renewable and storage investment. Only the most ambitious configurations for supporting green investment avoid this outcome, but they do so by marginalizing the wholesale price signal itself, thereby requiring a commitment to a hybrid market paradigm that is in contradiction with the rationale of the proposed price intervention.