Economics III - Climate Policy Analysis (HJ)

Questions • How strong of a mitigation effort should we undertake NOW (say, 2007-2015)? • Need more information? What specifically? Path for Today • Structure of the assessment task – The handling of uncertainty – Representation of decision-making process – Areas of policy choice • Examples under Certainty – Benefit-cost analysis – Cost-effectiveness analysis – Tolerable windows analysis • Examples under Uncertainty – Probabilistic forecasts – Sequential decision Note: we get a bit ahead of ourselves here Certainty vs. Uncertainty • Assuming certainty – Once-and-for-all decision now • Near-term choice (e.g., Kyoto-type analyses) • Path over time (e.g., B/C, stabilization studies) • Considering uncertainty – Once-and-for-all decision now • Scenario analysis • With probability distributions – Sequential choice, with learning How important to include uncertainty? Representation of the DecisionMaker or Process • Single decision-maker • Multiple decision-makers and gaming behavior • Negotiation among parties What is the value/limits of single-actor analysis? Areas of Policy Choice • Emissions control (what to do now?) Ã – Single decision-maker (global welfare) – [Multiple parties and negotiation] • Anticipatory adaptation • Actions to open options – R&D & technology push – “Architecture” of climate negotiations • Geo-engineering Examples under Certainty (1) B/C (2) What control action today? Who does what? (3) (4) CE TWA Climate target (Article 2) Benefit-Cost Analysis • Cost function & benefit relationship • Alternative applications – Calculate optimal path, unconstrained – Constrain by long-term target – Apply policy scenarios (e.g., burden sharing) • Difficult issues – Valuation & aggregation – Discounting – Institutional assumptions Example: Nordhaus DICE Model • Growth, emissions, and ΔT – Like HW #3 • Climate change effects – A damage function of form in last lecture • Forward-looking, optimizing model • Policy assumptions – Optimal path (by their valuations) – Stabilize concentrations at 2xCO2 – Hold ΔT to 2.5°C – Stabilize emissions at 1990 levels (E90) Efficient Policies Efficient Climate-Change Policies 70 Control Rates - % of baseline 60 50 E90 40 2XCO2 30 T < 2.5 20 Optimal 10 0 1995 2005 2015 2025 2035 2045 2055 2065 2075 2085 2095 -10 Emission control rates: Alternative policies Optimal E90 2xCO2 T < 2.5 deg C 2105 Social Cost of Carbon 100 2XCO2 E90 90 T<2.5 80 Carbon taxes (1990 $ per ton) 70 60 Optimal 50 40 30 20 10 0 1995 2005 2015 2025 2035 2045 2055 2065 2075 2085 Carbon taxes: Alternative policies Optimal E90 2xCO2 T < 2.5 deg C 2095 2105 Insights/Evaluation? • What think of the analysis? • Insights gained? – About paths of stringency? – Other? • What assumptions dominate? – – • What is missing? – – Cost-Effectiveness Analysis • No direct benefit estimate – Least-cost path in stabilization studies – Examples: CCSP (previous lecture) & HW #3 • Quantify what, where & when flexibility • Input to meta benefit-cost analysis – Combine with benefits of stabilization level – Example: Stern Review • Difficult issues – Aggregation – Discounting – Institutional assumptions Tolerable Windows • No explicit benefit function – Represented in form of constraints • No explicit cost function – Represented by some limit on effort • Question: what must we do to preserve the option of some future climate state? – Capable of multiple attributes Sequence of Windows Climate Space 0.20 0.15 0.10 0.05 460 440 420 400 300 360 340 0.00 15.0 15.5 16.0 o Composition Space 480 CO2 Concentration [ppm] Rate of Change (oC/dec) 0.25 16.5 0 50 17.0 100 150 200 Years After 1995 Temperature ( C) Emission Space (2) 15 1000 Emissions/Year [Gt C/a] Cumulative Emissions [Gt C] Emission Space (1) 800 600 400 200 0 10 5 0 0 50 100 Years After 1995 150 200 0 50 100 Years After 1995 150 200 Distribution of Plant Communities 450 Tropical Rainforest Temperate Rainforest Temperate Forest Taiga Tundra Grassland Desert Tropical Seasonal Forest 400 350 Precipitation (cm) 300 250 200 150 100 50 30 25 20 15 10 5 0 -5 -10 -15 Mean Annual Temperature oC Tropical Subtropical Warm Temperate Cold Temperate Arctic Alpine Figure 2. Corridors for energy-related CO2 emissionsCorridors for energy-related CO2 emissions (a) Variation of the impact constraint 24 Maximum global ecosystem transformation 50% 45% 40% 35% 30% Maximum regional income loss: 2.0% CO2 emissions ( Gt C/yr) 20 16 Inner structure (for 35% limit) 12 Max. 2035 Min. 2035 Max. 2065 Min. 2065 Max. 2100 Min. 2100 8 4 0 2000 2020 2040 Year 2060 2080 2100 Cost-effective emission path Insights/Evaluation? • What think of the analysis? • Insights gained? – – • What assumptions dominate? – Structure of solution algorithm – • What is missing? – – Assessing an Atmospheric Target Under Uncertainty What control action today? Meta B/C (2) Climate target (Article 2) What would we gain with stabilization & 550 ppm? Benefit-Cost Under Uncertainty (1) What control action today? B/C Climate target (Article 2) The “Wait to Learn” Debate • How should the ability to learn (reduce uncertainties) influence actions today? (EOS debate) – Do less if expect to learn? – Do more? • Because of what? – Stock nature the problem (irreversibility) – Degree and origin of the uncertainty? – How soon learning comes? – Discount rate? Alternative Decision Formulations Learn Do more Do more Do less Do less Do more Now Learn Learn Do more Do less Do less Do more Do more Do less Do less Future Learn “Act then Learn” Example Do more Do less • Damage – Low – Moderate – High Learn Do more Do less Learn X Emissions Unlimited 20% below 1990 50% below 1990 Hedging Under Uncertainty 3.5 Unlimited Billion Tons of Carbon 3.0 2.5 2.0 1.5 1.0 20% Emissions Cut 0.5 0 1990 50% Emissions Cut 2000 2010 Act, then Learn 2020 2030 2040 Learn, then Act 2050 Missing: Institutions and Path Dependency • Normal assumptions in studies of choice over time – Optimal paths will be followed when information becomes available – Wait and decide later • Alternative hypotheses about capability? – – Ongoing Research • Upper tail the distribution of outcomes – Missing (extreme) events • Methodology – Elicitation of parameter PDFs – Cascading uncertainties through models of several stages of the climate issue • The real (sequential) decision problem – Partial learning – Institutions and path dependency – Capturing risk aversion (precaution) – Multiple players & “who does what?” • Lay communication Final Thoughts • At best, gain rough insight to today’s decision – Damage functions are inadequate to the task – Necessary simplification of choices – Thus far: single decision-maker model, or very simple decision theory representations • Much work needed to do better, even for “expert” understanding • Lay audiences deserve our sympathy