Carbon Policy Landscape: How Price, Subsidies and Enforcement Shape Emissions, CO2 and Warming

Set a carbon price, a subsidy and how well it is enforced, then watch 75 years of emissions, CO2 and warming unfold over the landscape.

2D map

Policy presets

Playback

Policy levers

Camera

Quality

Model checks

The science

What this is. A teaching model of one stylised economy sized like the whole world (about 40 GtCO2 per year in 2025) under a single global policy. Every number is illustrative: it shows how the levers interact, it is not a forecast and not a calibrated integrated-assessment model. The landscape is a picture of the model state, not a separate simulation. The model is recomputed from 2025 to 2100 whenever you move a slider; the time-lapse then plays the result.

1. Carbon price and demand response (constant elasticity). Each of five sectors i has baseline emissions E0,i, a price scale p0,i and an elasticity εi: Ei(p) = E0,i (1 + p/p0,i)−εi. The price rises as p(t) = p2025 (1 + r)t. Power is the most price-responsive sector (ε = 0.55) and agriculture and land the least (0.12); these values are round teaching numbers, not estimates.

2. Compliance and enforcement. Only the share c that complies faces the price: Ei = E0,i egt [ c (1 + p/p0,i)−εi + (1 − c) ] (1 − φi σ), with baseline growth g = 0.5 % per year. The non-compliant share carries on as before and pays nothing, so revenue is p Σ c E0,i egt Di (1 − φi σ) in billion dollars per year (Gt × $/t).

3. Clean-power substitution with a learning curve. Investors compare the cost of fossil power including the carbon charge, Cf = 55 + 0.7 c p ($/MWh, 0.7 tCO2/MWh), with clean power after subsidy, Cc (1 − s). The target clean share is logistic, f* = 1 / (1 + e−(Cf − Cc(1−s))/w) with w = 10 $/MWh, and the installed share follows it with a capital turnover time τ = 25 years and no early retirement: df/dt = max(0, f* − f)/τ. Clean cost falls with cumulative deployment X (Wright's law): Cc = Cc,0 X−b, b = −log2(1 − LR), so every doubling cuts the cost by the learning rate LR. X is the installed share relative to 2025 times a 2 % per year spillover from the rest of the world, which is why clean power slowly gets cheaper even with no policy. Each sector can only be cleaned up to its addressable share φi (95 % power, 60 % buildings, 45 % industry, 35 % transport, 10 % agriculture): σ = (f − f0)/(1 − f0), f0 = 20 %.

4. Money. Abatement (resource) cost of the demand response is the area between the demand curve and the price: ∫0p p'|dE/dp'| dp' = E0 [ p0/(1−ε) ((1 + p/p0)1−ε − 1) − p (1 + p/p0)−ε ]. The clean build-out premium is the extra cost of clean over fossil power on the new share, and can turn negative (savings) once clean power is cheaper. The subsidy bill is s Cc Q f with Q = 30 000 TWh per year. Revenue not returned to households (the slider) pays the subsidy; the government balance is (1 − d) Revenue − subsidy bill, and a negative value means the package is not self-financing. Costs are shown against a GDP of 110 000 billion dollars growing 2 % per year. Distributional effects, macro feedbacks, innovation spillovers and carbon leakage are not modelled.

5. From emissions to warming. Cumulative emissions Ec are integrated with RK4 (step 0.25 yr). CO2 in the air: C = 424 + AF · Ec / 7.79 ppm, with a constant airborne fraction AF = 0.45 (the real fraction varies with time and the state of the oceans and land) and 7.79 GtCO2 per ppm (2.124 GtC × 44/12). Warming uses the transient climate response to cumulative emissions: T = 1.3 + 0.45 Ec / 1000 °C, the central IPCC AR6 value (likely range 0.27 to 0.63 °C per 1000 GtCO2), starting from about 1.3 °C above 1850-1900. Only CO2 is counted: no methane, aerosols, tipping points or carbon-cycle feedbacks, and the TCRE relationship is only roughly linear. Treat the warming figures as illustrative.

Reading the scene. Dark plumes grow with emissions from power, industry, homes and traffic (white steam from the cooling towers is water vapour); wind turbines and solar panels appear as the clean share rises; teal cars are electric; the sky hazes as CO2 builds up. Every constant (elasticities, costs, learning rate, turnover, subsidy bill) is a round teaching value chosen for plausibility, not a measured one: please verify against a real source before quoting any of them.

Checks on this page. Seven independent checks run when the page loads (green boxes): the numerical log-log slope of every sector's demand curve equals −ε; the closed-form abatement cost matches a trapezoid integral of p|dE/dp|; the learning curve halves per doubling exactly as defined; the RK4 substitution ODE matches its exact exponential solution; cumulative emissions agree with an independent Simpson sum and with a five times finer step; the ppm and warming chain agrees with a separate unit conversion; and cumulative emissions fall monotonically with price and with compliance across a sweep.

Sources. IPCC AR6 WG1 (TCRE, carbon budgets), Global Carbon Budget (airborne fraction, GtC per ppm), Wright 1936 and Rubin et al. 2015 (learning curves, typically 10 to 25 % per doubling for power technologies), Nordhaus-style cost-benefit and carbon-tax literature (revenue recycling). Citations written from memory: please verify before quoting. Not affiliated with any government or agency.

Year, emissions, CO2, warming-
Loading model...
Loading Blender model...
Emissions by sector against the no-policy total
Warming above 1850-1900 (illustrative TCRE)
Fossil power with carbon cost against clean power after subsidy
Made with 🟠 Blender · Three.js