Climate-Macro ABM Model
How do carbon emissions, climate damages, green innovation, and carbon pricing interact with heterogeneous firms, banks, and households to produce transition risk, stranded assets, and macroeconomic outcomes that integrated assessment models miss?
Understand the mathematics
Follow the same equations and assumptions one section at a time or in a continuous reading.
Hover or focus a symbol defined in this model to find its exact matches. Select it to keep the highlight; Escape clears the selection.
Odd+D Protocol: Model Specification
The ODD+D protocol (Grimm et al. 2006, 2010; Muller et al. 2013) applied to the climate-macro ABM family. The specification draws primarily on the DSK model (Lamperti et al. 2018, 2020) as the most fully documented member of this class, with notes on where EIRIN and FTT diverge.
PURPOSE: The model investigates how heterogeneous firms with different carbon intensities, banks with differential exposure to carbon-intensive borrowers, and endogenous green innovation interact to produce transition risk, stranded assets, and macroeconomic outcomes under various carbon pricing and green industrial policy designs. The target audience is central banks evaluating climate-related financial stability risks and policymakers designing carbon tax and green subsidy packages.
ENTITIES, STATE VARIABLES, AND SCALES: The model contains five entity types. Consumption goods firms ( to in research runs) carry: labor productivity, carbon intensity, capital stock, debt, equity, inventories, and market share. Capital goods firms ( to ) carry: green technology level, machine carbon intensity, R&D budget, and client list. Households ( to ) carry: income, wealth, employment status, and consumption budget. Banks ( to ) carry: loan portfolio with carbon exposure decomposition, capital ratio, and lending rule parameters. The government carries: carbon tax rate, subsidy budget, and fiscal balance. The climate module carries: CO2 concentration, temperature anomaly, and damage distribution parameters. Temporal scale is quarterly periods over 80 to 120 quarters (20 to 30 years).
PROCESS OVERVIEW AND SCHEDULING: Each period proceeds in fixed order: (1) climate module updates CO2 and temperature from prior-period emissions; (2) climate damage shocks are drawn and applied to firm capital and productivity; (3) capital goods firms conduct R&D and potentially improve green technology; (4) consumption goods firms choose capital investment (selecting machines from capital goods firms) and set production plans; (5) the goods market clears (firms produce, households consume); (6) the labor market adjusts (hiring and firing); (7) firms calculate profits, service debt, and potentially default; (8) banks process defaults, update capital, adjust lending standards; (9) the government collects carbon tax, distributes subsidies, runs fiscal policy; (10) aggregate statistics and period emissions are recorded. The ordering ensures that climate damages are current-period information when firms make production decisions.
CO2 concentration update. Only the EXCESS over the pre-industrial level 280 ppm decays at rate (natural carbon sinks); decaying the entire stock would send CO2 to zero under zero emissions and the anomaly to minus infinity. The stock rises with last period's emissions, scaled by . Timing is uniform throughout the file: period 's concentration uses period 's emissions, matching the schedule in which the climate module runs first.
Temperature relaxes toward its equilibrium target at speed . The forcing coefficient is , NOT : is the equilibrium climate sensitivity, warming at a CO2 DOUBLING, so the coefficient on must be at for a doubling to deliver exactly 3.0 °C. The right-hand side uses explicitly, as every other rendering in this file does. This is a linearized energy balance model whose doubling response is calibrated to the IPCC AR6 equilibrium climate sensitivity range.
Two-stage damage, and the only damage specification used anywhere in this file: first a Bernoulli draw determines whether firm is hit at all (probability increasing quadratically with temperature, scaled by ); second the severity of the hit is drawn from a Beta whose mean shifts right with temperature. This produces heterogeneous and temperature-dependent damage across firms. It does NOT produce fat-tailed damage: a Beta is bounded on [0,1], so the worst outcome any firm can draw is total loss of the struck capital and there is no heavy tail. Weitzman-style tail risk would require replacing the severity distribution outright.
What the ABM adds beyond DICE
DICE maps emissions to temperature to a deterministic aggregate damage fraction of GDP. The ABM disaggregates this into firm-level stochastic shocks, which means: the same average damage produces different GDP outcomes depending on which firms are hit and how losses propagate through banks; damage variance and CONCENTRATION matter alongside the mean - a given average loss landing on a few borrowers of one bank does more harm than the same loss spread evenly, which is the ABM's actual amplification mechanism (the baseline severity draw is a bounded Beta, so this is a concentration effect, not a fat-tail effect); the transition path matters because firm heterogeneity in carbon intensity means a carbon tax creates winners and losers whose relative sizes change over time.
Stranded assets as an emergent phenomenon
Stranded assets in this model are not defined exogenously (e.g., 'X% of fossil reserves become unburnable'). They emerge endogenously: when a carbon tax makes high-carbon capital goods economically obsolete, the firms using them cannot cover costs, their market value drops, and the bank loans secured against them lose value. The timing and magnitude of stranding depend on the interaction between carbon tax ramp-up speed, green innovation rate, and bank portfolio composition - none of which can be computed from a single variable.
The financial amplification channel
Lamperti et al. (2019, Nature Climate Change) documented a 15-30% amplification of GDP losses through banking. The mechanism: climate damages cause firm defaults, bank capital absorbs losses, banks tighten credit, credit-constrained firms cut investment (including green investment), the economy becomes more fragile and more carbon-intensive, and the next round of damages hits harder. This positive feedback loop between climate damage and financial fragility is the model's most policy-relevant finding and is completely absent from IAMs.
Transition checks
Agent update
The aggregate is computed after individual states are updated. If the rule is nonlinear or stochastic, the aggregate can change abruptly even when the plotted line connects adjacent periods.
Each point is an aggregate after one model update. The connecting line helps the eye, but the law is discrete.
Do not read the line as a continuous differential equation unless one is stated.
Cutoff and contagion
The indicator changes value discontinuously at . Aggregates inherit that discontinuity when enough agents sit near the cutoff.
A cutoff creates a kink or jump because the update rule changes at the threshold.
Do not force a smooth explanation onto a chart section driven by threshold crossing.
Variable timing
Agent state
Agent states update at dates. Abrupt aggregate movement can come from many small state changes applied at once.
Aggregate readout
The aggregate can jump when the distribution changes shape or many agents cross a boundary.
Threshold indicator
The indicator turns on at the cutoff. That creates a discontinuity in the micro rule and possibly the aggregate.