Integrated assessment modelling

Based on Wikipedia: Integrated assessment modelling

The Impossible Models That Shape Our Climate Future

Here's a question that should keep you up at night: How do you predict the future of human civilization?

Not just the weather next week, or stock prices next quarter, but the intertwined fate of eight billion people, their economies, their energy systems, their farmland, their cities—all of it dancing with an atmosphere that's warming in ways we've never experienced. This is the challenge that integrated assessment models attempt to solve. They are, without exaggeration, among the most ambitious intellectual constructions humans have ever attempted. They are also, by necessity, deeply flawed.

And yet these models shape trillion-dollar policy decisions. They've influenced the Paris Agreement. They calculate numbers that governments use to set carbon taxes and evaluate regulations. Understanding what these models can and cannot do isn't just an academic exercise—it's essential for anyone trying to make sense of climate policy debates.

What Integrated Assessment Models Actually Do

Integrated assessment modelling—often abbreviated as IAM, though we'll spell it out—tries to accomplish something no single scientific discipline can do alone: link human society with the physical Earth system in a single mathematical framework.

Think about what that requires. You need economics—how markets work, how people make decisions, how industries grow and decline. You need climate science—how greenhouse gases trap heat, how the atmosphere and oceans interact, how ice sheets respond to warming. But that's just the beginning.

The most comprehensive models also incorporate energy systems (how we generate and use power), land use (farming, forests, cities), agriculture (crop yields under different conditions), infrastructure (roads, buildings, power grids), and even governance (how policies get implemented). Some venture into education, health, and conflict.

The word "integrated" captures this sprawling ambition. These models try to span multiple academic disciplines that rarely talk to each other. An economist studying carbon taxes and a climate scientist modeling ice sheet dynamics speak different languages, use different methods, care about different timescales. Integrated assessment models force them into conversation.

The word "assessment" points to the purpose: informing policy decisions. These aren't pure scientific curiosity. They exist to answer questions like "What happens if we limit warming to 1.5 degrees Celsius?" or "How much should we charge for carbon emissions?"

Two Fundamentally Different Species

Not all integrated assessment models work the same way. There are two fundamentally different species, and confusing them leads to endless misunderstanding.

The first type we might call process-based models. These are intricate simulations that try to represent how the world actually works—sector by sector, region by region, year by year. They trace energy flowing through the economy, emissions rising into the atmosphere, temperature changes feeding back into agricultural productivity. The Intergovernmental Panel on Climate Change, or IPCC, relies heavily on these models to explore different pathways for meeting climate targets.

If you've heard of the "1.5 degree pathway" from the Paris Agreement, process-based integrated assessment models helped define what that pathway might look like. How fast would we need to decarbonize electricity? Which technologies would need to scale up, and by when? What role might carbon capture play? These models generate detailed scenarios—not predictions, but internally consistent stories about how the future might unfold.

Notable frameworks in this category have names like IMAGE, MESSAGEix, GCAM, and REMIND-MAgPIE. Each has different strengths and assumptions, but they share the ambition of simulating complex processes in granular detail.

The second species is quite different: cost-benefit models. Rather than simulating detailed processes, these models aggregate everything into simplified equations focused on total costs. What's the economic cost of climate change damages? What's the cost of reducing emissions? Where's the optimal balance?

These models serve a specific purpose: calculating something called the social cost of carbon. This is the marginal social cost—in dollars—of emitting one additional ton of carbon dioxide into the atmosphere. If burning a gallon of gasoline releases about 9 kilograms of carbon dioxide, and the social cost of carbon is $50 per ton, then that gallon of gas imposes about 45 cents of damage on society beyond what the driver pays at the pump.

The social cost of carbon matters enormously. The United States government uses it to evaluate regulations. If a proposed rule would reduce emissions, the benefits of that reduction get calculated using the social cost of carbon. Models like DICE, PAGE, and FUND have been used by the U.S. Interagency Working Group to generate these numbers.

The Market Failure at the Heart of Climate Change

Why do we need to calculate the social cost of carbon at all? The answer reveals a fundamental problem in how markets handle climate change.

When you buy gasoline, the price reflects the cost of extracting oil, refining it, and delivering it to the station. But it doesn't include the cost of the climate damage from burning it. Economists call this a negative externality—a cost imposed on others that isn't reflected in the market price.

Externalities cause market failures. When the price is wrong, people make decisions that don't account for full costs. We burn more fossil fuels than we would if their price reflected climate damage. Markets, left alone, won't solve this problem.

A carbon tax is one way to correct this market failure. Make emitters pay for the damage they cause, and the price signal ripples through the economy. But to set the tax correctly, you need to know the cost—hence the social cost of carbon.

There's a catch, though. The social cost of carbon is extraordinarily uncertain. How much economic damage will a degree of warming cause? How do we value damages that occur decades from now? How do we handle the possibility of catastrophic tipping points?

These questions don't have clean answers. Estimates of the social cost of carbon range from under $20 per ton to over $200 per ton, depending on assumptions. That's not a minor uncertainty—it's the difference between modest policy adjustments and a wholesale restructuring of the global economy.

The Critique That Won't Go Away

Here's where things get uncomfortable. These models, which inform decisions worth trillions of dollars, have been severely criticized.

The criticisms cut deep. Some economists argue that the models have systematically underestimated the benefits of aggressive climate action while overestimating its costs. The assumptions baked into the models—about how economies grow, how damages scale with temperature, how we should value future generations—aren't neutral technical choices. They're philosophical positions with massive policy implications.

Nicholas Stern, a prominent economist who authored an influential 2006 report on climate economics, argued in 2021 that existing integrated assessment models are "inherently unable to capture the economic realities of the climate crisis under its current state of rapid progress." The world is changing faster than the models assume.

Another critique draws on dynamical systems theory—the mathematical study of how complex systems evolve. This perspective holds that the future isn't a single pathway to be optimized, but a vast space of possibilities, many of which we can't even imagine from where we stand today. The models assume we can know the shape of future possibility space. We can't.

This type of uncertainty has been called "radical" or "fundamental" uncertainty. It's not just that we don't know which outcome will occur—we don't even know all the possible outcomes. Under such conditions, optimization models may be asking the wrong question entirely.

Critics argue that the models "create a perception of knowledge and precision that is illusory, and can fool policy-makers into thinking that the forecasts the models generate have some kind of scientific legitimacy." Strong words. But the critics aren't cranks—they're serious researchers pointing to real problems.

Why We Use Them Anyway

So why do policymakers keep using these models? Are they simply fooling themselves?

The honest answer is more nuanced. Even deeply flawed models can provide useful insights, as long as we interpret them correctly.

Process-based models help us think through scenarios. What would a world running on renewable energy actually look like? What are the bottlenecks? What infrastructure would we need? Even if the model can't predict which scenario will occur, it can illuminate what different futures would require.

Cost-benefit models, despite their limitations, force us to think systematically about trade-offs. They make assumptions explicit rather than leaving them implicit. A model that estimates the social cost of carbon at $50 per ton is making claims about discount rates, damage functions, and climate sensitivity that can be examined and debated. An intuition that "carbon should cost more" is harder to scrutinize.

The alternative to flawed models isn't perfect models—it's no models. And no models means policy made on gut instinct, lobbying power, and ideology. At least models provide a structured way to organize evidence and surface disagreements.

The Possibility Space

In 2021, the integrated assessment modeling community conducted a fascinating self-examination. They looked at what they called the "possibility space"—the range of futures the models could explore—and asked what was missing.

The findings were revealing. Most scenarios assumed continued economic growth. But what about post-growth scenarios—futures where societies prioritize well-being over gross domestic product? What about scenarios with radically different governance structures, or transformative social movements, or technological breakthroughs we can't yet imagine?

The models, it turned out, were exploring only a narrow slice of what might be possible. The scenarios that received the most attention from researchers weren't necessarily the most likely or most important—they were the ones that fit most comfortably within existing modeling frameworks.

This is a deep problem. If the models can only explore certain types of futures, they may inadvertently constrain our imagination about what's achievable. A scenario that's difficult to model isn't the same as a scenario that's unlikely to occur. The map isn't the territory.

Beyond Climate

Interestingly, integrated assessment models aren't used only for climate change. The underlying approach—linking economic, social, and physical systems in a single framework—has been applied to other questions.

Researchers have used these methods to analyze patterns of conflict, exploring how resource scarcity, economic stress, and governance failures might interact. They've modeled progress toward the Sustainable Development Goals, the United Nations' ambitious 17 targets for human flourishing. They've examined food security, tracing connections between agriculture, climate, trade, and nutrition.

In Africa, integrated models have been used to analyze trends across issue areas—health, education, infrastructure, governance—simultaneously rather than in isolation. This matters because these domains interact in complex ways. Education affects health outcomes, which affect economic productivity, which affects government revenue available for education. Treating each issue separately misses these feedback loops.

The Equilibrium Assumption

Many integrated assessment models make a powerful assumption: that economies tend toward equilibrium. Markets clear. Supply meets demand. Prices adjust until everything balances.

This assumption has a long pedigree in economics, but it's controversial when applied to the climate problem. Climate change happens over decades and centuries. Are markets really in equilibrium over such timescales? Do people make rational decisions about risks that won't materialize until their grandchildren are old?

Some models reject the equilibrium assumption entirely. Non-equilibrium models use approaches like econometrics—statistical analysis of historical patterns—or evolutionary economics, which treats the economy as a complex adaptive system rather than a machine tending toward balance.

Agent-based models take a different approach still. Rather than modeling aggregate quantities like "the economy" or "consumers," they simulate individual agents—people, firms, governments—each following rules that may or may not be rational. The aggregate behavior emerges from millions of individual interactions. These models can produce surprising results, including abrupt transitions and persistent disequilibrium, that equilibrium models miss.

What It All Means

If you've made it this far, you might be feeling dizzy. The models are ambitious but flawed. They're essential but misleading. They inform policy but can't predict the future.

Welcome to the reality of decision-making under deep uncertainty.

The honest conclusion isn't that integrated assessment models are useless or that we should abandon them. It's that we need to use them wisely, with clear eyes about what they can and cannot do.

They can illuminate trade-offs and help us think through scenarios. They can make assumptions explicit and force systematic reasoning. They can identify bottlenecks and highlight interactions between systems.

They cannot predict the future. They cannot tell us the "right" climate policy. They cannot resolve fundamental disagreements about values—how much we should sacrifice today for future generations, how to weigh certain costs against uncertain catastrophes, whose interests count most in a global problem.

These are questions that models can inform but not answer. The answers come from politics, ethics, and collective choice. The models are tools. What we build with them is up to us.

In a world where climate change is already reshaping economies and ecosystems, where policy decisions made today will echo for centuries, that responsibility weighs heavy. The models won't save us. But neither can we navigate this challenge without them.