Ask the Scholar

Document scope · 1 page
doc
Scholar
Ask about this object, its catalog metadata, its source description, or the page inventory. For page-specific OCR and visual context, open one of the page chats.

Scholar Source Context

Document identity
localId
225358440
label
Climate Change - Trading Models [1]
core
doc
dtoType
document
pageCount
1
Source metadata
Source extras
naId
225358440
levelOfDescription
fileUnit
otherTitles
42-t-4739519-20171095F-011-010-2019
recordType
description
ocrSource
nara-archive
Single page context
seq
1
pageIndex
0
type
document
mediaId
bf495501d0d10b9d
ocrText
FOIA Number: 2017-1095-F FOIA MARKER This is not a textual record. This is used as an administrative marker by the William J. Clinton Presidential Library Staff. Collection/Record Group: Clinton Presidential Records Subgroup/Office of Origin: Council of Economic Advisers Series/Staff Member: Judson Jaffe Subseries: OA/ID Number: 20747 FolderID: Folder Title: Climate Change - Trading Models [1] Stack: Row: Section: Shelf: Position: S 20 6 1 3 MEMORANDUM To: Janet Yellen From: Joe Aldy and Randy Lutter Date: 10/8/97 Subject: Sensitivity of permit prices to alternative economic assumptions How sensitive are estimated carbon permit prices to alternative assumptions about GDP growth, autonomous energy efficiency improvement, and the opportunities for substitution across technologies and fuels? A preliminary assessment of the effects of varying these assumptions indicates that permit prices are unlikely to fall significantly. Since fuel switching is sensitive to elasticities embedded in models, we conducted our analysis with energy and carbon outputs from the Markal Macro and SGM models. We find that for the permit price to be $30/ton in the SGM model, AEEI would have to equal GDP growth. (This implies zero growth in energy use over the decade.) For the permit price to be $30 in the Markal Macro model, AEEI would have to exceed GDP growth, implying that energy use would have to fall over this period. The only recent decade when energy consumption did not grow was from 1972 to 1982. Methodology We assume that the price of carbon varies linearly with the emissions reduction required under a 1990 in 2010 stabilization case relative to a no policy baseline. CHICE where Pₛ is the forecast price under stabilization under our approach, Pₘ is the model output forecast price under stabilization, Cₛ is the change in carbon emissions required to achieve 1990 level in 2010 from a business as usual level (specified below), and Cₘ is the reduction in carbon relative to the models' estimated business as usual. Emissions reductions can be written as S C/F the product of a fuel switching factor, which is defined as the change in carbon under a stabilization policy divided by the change in energy under this policy (both relative to a business as usual projection for 2010) from a model, multiplied by the percentage change in energy consumption. (Note that since the fuel switching parameter is driven by the price of carbon, we cannot vary it exogenously.) The change in total energy use between the stabilization case and business as usual in 2010 is determined by GDP growth and autonomous energy efficiency improvement inputs. As a starting point, we use the results from SGM and Markal Macro with the IAT assumptions (AEEI = 1.25%, GDP growth of 2.06% and 1.91%) and permit prices of $81/ton and $145/ton in 2010. We use these to calibrate our spreadsheet. Note that our calibrations result in price estimates that differ from the model results by 5.8% and 2.6% -- this likely is a function of using rounded averages for GDP and AEEI. SGM generates a fuel switching factor of 1.5, while Markal Macro assumes more fuel switching, so its factor is 1.64. We then calculated a set of permit prices by varying GDP and AEEI over the 1995-2010 period. We varied annual GDP growth by about 1.3% and 3% and AEEI by 0.7% and 1.75%. Note that the most recent GDP growth rate projection is about 2.4%, and a survey of energy economists indicates that AEEI likely falls between 0.5% and 0.7% (EIA employs a 0.9% energy efficiency improvement). Results The assessment of the permit prices with varying input assumptions indicates that permit prices can vary significantly with the inputs. However, given that the IAT employed very optimistic AEEI (1.25%) and moderate economic growth (~2%), the variations of inputs imply that permit prices could easily be above the IAT forecast results. For example, assuming an AEEI of 0.9% (consistent with the Energy Information Administration) and the present projected GDP growth rate of 2.4%, the computed permit price with the SGM derived measure would be about $130/ton while the Markal Macro derived measure would be about $250/ton. The chart on the next page presents the estimated permit prices given the ranges of GDP growth and AEEI considered. Low permit prices would require very slow growth and very optimistic AEEI. For example, assume that over the next 13 years, the economy experiences 8 years of zero growth (e.g., a prolonged recession) and five years of moderate growth (e.g., 2% per year). During this period, assume that autonomous energy efficiency improvements occur at an annual rate of 0.9%. Under this scenario, 2010 energy consumption would be roughly equal to 1995 energy consumption. The price of a carbon permit in 2010 would still be about $30/ton with the SGM derived measure, and nearly $60/ton with the Markal Macro measure. Our sensitivity analysis indicates that prices can vary, but are more likely to vary upward relative to the IAT results. Note however that we only varied GDP growth and AEEI. The nature of the supply curve for emissions reductions also can affect the permit price. Modifying other assumptions may generate different results. 1 As a check on our proportionality assumption, we did a second calibration of SGM using an AEEI of 1.0%, GDP growth of 2.0%, a fuel switching factor of 1.39, and a permit price of $108. This results in a predicted price of $121, or 12% from the actual model result. Our assumption does not appear to be far off, but with higher emissions reductions, our approach generates permit prices slightly larger than the model estimates. With smaller emissions reductions, the error appears to be small. Note the attached chart that illustrates the marginal abatement cost curve for emissions reductions in 2010. Estimated Permit Prices Under Various GDP Growth and AEEI Assumptions Derived from: GDP Growth Rate AEEI Permit Price ($/ton) SGM* 2.06% 1.25% $86 SGM, Best Estimate 2.4%¹ 0.9%¹ $128 SGM 1.25%² 0.7% $69 SGM 1.25%² 0.9% $55 SGM 1.25%² 1.25% $31³ SGM 1.25%² 1.75% $0³ Markal Macro* 1.91% 1.25% $149 Markal Macro, Best Estimate 2.4%¹ 0.9%¹ $250 Markal Macro 1.25%² 0.7% $135 Markal Macro 1.25%² 0.9% $108 Markal Macro 1.25%² 1.25% $60³ Markal Macro 1.25%² 1.45% $31³ Markal Macro 1.25%² 1.75% $0³ * Calibration estimates: SGM model result is $81/ton, Markal Macro model result is $145/ton. 1: The 2.4% GDP growth rate corresponds to the Administration's most recent projection of GDP growth. The 0.9% AEEI corresponds to the Energy Information Administration's projection of AEEI over the next 20 years. 2: A growth rate average of 1.25% between now and 2010 would imply that during six of the next thirteen years, the nation would experience a GDP growth rate of zero (e.g., a prolonged recession) and annual GDP growth would average, during the remaining seven years, about 2%. 3: These estimates of permit prices assume that energy use remains flat or falls between now and 2010. An energy trend of this nature, over a period as long as a decade, has only occurred during the 1970s and early 1980s as a result of the OPEC oil shocks. Carbon Emissions Reductions Marginal Abatement Cost Curve, 2010 Carbon Price ($/ton) 200 150 100 Emissions reductions required 50 for stabilization at 1990 level 0 0 100 200 300 400 500 600 Emissions Reductions (MMTCE) Model: SGM 9/18/97 Outputs ID Scenario Model Modelers Target Timetable Trading Permit Revenue Burden BAU Paper Tons AEEI Ramp-up Time Path PDV (5%; GDP in 2010 Allocation Recycling Sharing Emissions or Corre- 2000-2050) (deviation Path sponding Foregone from BAU) Assumptio Reductions Consumption a SGM1 BAU SGM, Battelle n/a n/a n/a n/a n/a n/a IAT n/a 1.0 n/a -->2050, $121,650 billion $9,185 billion MAGICC -->2100 (BAU (BAU GDP) (climate) consumption) SGM9 -10% of SGM, Battelle -10% 1990 stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 no -->2050, $485 billion $9,155 billion 1990 in MAGICC emissions 2010 part. -->2100 (-$30 billion) 2010 level (climate) SGM8 -10% of SGM, Battelle -10% 1990 stabilize in domestic auction lump-sum no LDC IAT PT 1.0 no -->2050, $980 billion $9,149 billion 1990 in MAGICC emissions 2010 only part. -->2100 (-$36 billion) 2010 level (climate) SGM10 -10% of SGM, Battelle -10% 1990 stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 no -->2050, $50 billion $9,171 billion 1990 in MAGICC emissions 2010 part. -->2100 (-$14 billion) 2010 level (climate) SGM36 1990 in SGM, Battelle 1990 stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 yes -->2050, $195 billion $9,179 billion 2020 MAGICC emissions 2020 part. -->2100 (-$6 billion) level (climate) SGM39 1990 in SGM, Battelle 1990 stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $230 billion $9,172 billion 2010 MAGICC emissions 2010 part. -->2100 (-$13 billion) level (climate) SGM3 1990 in SGM, Battelle 1990 stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 no -->2050, $225 billion $9,172 billion 2010 MAGICC emissions 2010 part. -->2100 (-$13 billion) level (climate) SGM18 1990 in SGM, Battelle 1990 stabilize in Annex I auction lump-sum LDC TAT PT 1.0 in 2005 -->2050, $255 billion $9,172 billion 2010 MAGICC emissions 2010 stabilizes -->2100 (-$13 billion) level at 2030 in (climate) 2030 SGM, Battelle 1990 stabilize in Annex I auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $260 billion $9,172 billion SGM21 1990 in 2010 MAGICC emissions 2010 BAU to -->2100 (-$13 billion) level 2030, (climate) equal per capita in 2050 SGM17 1990 in SGM, Battelle 1990 stabilize in domestic auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $740 billion $9,164 billion 2010 MAGICC emissions 2010 only stabilizes -->2100 (-$21 billion) level at 2030 in (climate) 2030 no LDC IAT PT 1.0 no -->2050, $665 billion $9,165 billion SGM2 1990 in SGM, Battelle 1990 stabilize in domestic auction lump-sum 2010 MAGICC emissions 2010 only part. -->2100 (-$20 billion) level (climate) SGM38 1990 in SGM, Battelle 1990 stabilize in domestic auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $740 bilion $9,164 billion 2010 MAGICC emissions 2010 only part. -->2100 (-$21 billion) level (climate) PT 1.0 in 2005 -->2050, $740 billion $9,164 billion SGM20 1990 in SGM, Battelle 1990 stabilize in domestic auction lump-sum LDC IAT 2010 MAGICC emissions 2010 only BAU to -->2100 (-$21 billion) level 2030, (climate) equal per capita in 2050 SGM40 1990 in SGM, Battelle 1990 stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $5 billion $9,180 billion 2010 MAGICC emissions 2010 part. -->2100 (-$5 billion) level (climate) 9/18/97 ID Scenario Model Modelers Target Timetable Trading Permit Revenue Burden BAU Paper Tons AEEI Ramp-up Time Path PDV (5%; GDP in 2010 Allocation Recycling Sharing Emissions or Corre- 2000-2050) (deviation Path sponding Foregone from BAU) Assumptio Reductions Consumption n SGM4 1990 in SGM, Battelle 1990 stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 no -->2050, $5 billion $9,180 billion 2010 MAGICC emissions 2010 part. -->2100 (-$5 billion) level (climate) SGM22 1990 in SGM, Battelle 1990 stabilize in worldwide auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $25 billion $9,180 billion 2010 MAGICC emissions 2010 BAU to -->2100 (-$5 billion) level 2030, (climate) equal per capita in 2050 SGM19 1990 in SGM, Battelle 1990 stabilize in worldwide auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $20 billion $9,180 billion 2010 MAGICC emissions 2010 stabilizes -->2100 (-$5 billion) level at 2030 in (climate) 2030 SGM35 1990 in SGM, Battelle 1990 stabilize in domestic auction lump-sum no LDC IAT PT 1.0 yes -->2050, $660 billion $9,172 billion 2020 MAGICC emissions 2020 only part. -->2100 (-$13 billion) level (climate) SGM37 1990'in SGM, Battelle 1990 stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 yes -->2050, -$I billion $9,183 billion 2020 MAGICC emissions 2020 part. -->2100 (-$2 billion) level (climate) SGM6 1995 in SGM, Battelle 1995 stabilize in Annex I auction lump-sum no LDC IAT CR 1.0 no -->2050, $380 billion $9,175 billion 2010 MAGICC emissions 2010 part. -->2100 (-$10 billion) level (climate) SGM5 1995 in SGM, Battelle 1995 stabilize in domestic auction lump-sum no LDC IAT CR 1.0 no -->2050, $425 billion $9,174 billion 2010 MAGICC emissions 2010 only part. -->2100 (-$11 billion) level (climate) SGM7 1995 in SGM, Battelle 1995 stabilize in worldwide auction lump-sum no LDC TAT CR 1.0 no -->2050, $30 billion $9,179 billion 2010 MAGICC emissions 2010 part. -->2100 (-$6 billion) level (climate) SGM33 1995 in SGM, Battelle 1995 stabilize in Annex I auction lump-sum no LDC IAT CR 1.0 yes -->2050, $410 billion $9,176 billion 2020 MAGICC emissions 2020 part. -->2100 (-$9 billion) level (climate) SGM32 1995 in SGM, Battelle 1995 stabilize in domestic auction lump-sum no LDC IAT CR 1.0 yes -->2050, $445 billion $9,176 billion 2020 MAGICC emissions 2020 only part. -->2100 (-$9 billion) level (climate) SGM34 1995 in SGM, Battelle 1995 stabilize in worldwide auction lump-sum no LDC IAT CR 1.0 yes -->2050, -$40 billion $9,180 billion 2020 MAGICC emissions 2020 part. -->2100 (-$5 billion) level (climate) SGM12 Peak in SGM, Battelle 2010 BAU stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 no -->2050, $55 billion $9, 185 billion 2015 MAGICC emissions in 2040 part. -->2100 ($0 billion) 2015; 1990 (climate) level in 2040 SGM30 Peak in SGM, Battelle 2010 BAU stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $75 billion $9,185 billion 2015 MAGICC emissions in 2040 part. -->2100 (0 billion) 2015; 1990 (climate) level in 2040 SGM24 Peak in SGM, Battelle 2010 BAU stabilize in Annex I auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $75 billion $9,185 billion 2015 MAGICC emissions in 2040 stabilizes -->2100 (0 billion) 2015; 1990 at 2030 in (climate) level in 2040 2030 9/18/97 ID Scenario Model Modelers Target Timetable Trading Permit Revenue Burden BAU Paper Tons AEEI Ramp-up Time Path PDV (5%; GDP in 2010 Allocation Recycling Sharing Emissions or Corre- 2000-2050) (deviation Path sponding Foregone from BAU) Assumptio Reductions Consumption n SGM27 Peak in SGM, Battelle 2010 BAU stabilize in Annex I auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $75 billion $9,185 billion 2015 MAGICC emissions in 2040 BAU to -->2100 (0 billion) 2015; 1990 2030, (climate) level in 2040 equal per capita in 2050 SGM11 Peak in SGM, Battelle 2010 BAU stabilize in domestic auction lump-sum no LDC IAT PT 1.0 no I-->2050, $235 billion $9,185 billion 2015 MAGICC emissions in 2040 only part. -->2100 ($0 billion) 2015; 1990 (climate) level in 2040 SGM29 Peak in SGM, Battelle 2010 BAU stabilize in domestic auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $270 billion $9, 84 billion 2015 MAGICC emissions in 2040 only part. -->2100 (-$1 billion) 2015; 1990 (climate) level in 2040 SGM26 Peak in SGM, Battelle 2010 BAU stabilize in domestic auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $270 billion $9, billion 2015 MAGICC emissions in 2040 only BAU to -->2100 (-$1 billion) 2015; 1990 2030, (climate) level in 2040 equal per capita in 2050 SGM23 Peak in SGM, Battelle 2010 BAU stabilize in domestic auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $270 billion $9,184 billion 2015 MAGICC emissions in 2040 only stabilizes -->2100 (-$1 billion) 2015; 1990 at 2030 in (climate) level in 2040 2030 SGM25 Peak in SGM, Battelle 2010 BAU stabilize in worldwide auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, -$3 billion $9,185 billion 2015 MAGICC emissions in 2040 stabilizes -->2100 (0 billion) 2015; 1990 at 2030 in (climate) level in 2040 2030 SGM31 Peak in SGM, Battelle 2010 BAU stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, -$5 billion $9,185 billion 2015 MAGICC emissions in 2040 part. -->2100 (0 billion) 2015; 1990 (climate) level in 2040 SGM13 Peak in SGM, Battelle 2010 BAU stabilize in worldwide auction lump-sum no LDC TAT PT 1.0 no -->2050, -$8 billion $9, 185 billion 2015 MAGICC emissions in 2040 part. -->2100 ($0 billion) 2015; 1990 (climate) level in 2040 SGM28 Peak in SGM, Battelle 2010 BAU stabilize in worldwide auction lump-sum LDC IAT PT 1.0 in 2005 -->2050, $1 billion $9,185 billion 2015 MAGICC emissions in 2040 BAU to -->2100 (0 billion) 2015; 1990 2030, (climate) level in 2040 equal per capita in 2050 SGM15 +10% of SGM, Battelle +10% 1990 stabilize in Annex I auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $95 billion $9,181 billion 1990 in MAGICC emissions 2010 part. -->2100 (-$4 billion) 2010 level (climate) SGM14 +10% of SGM, Battelle +10% 1990 stabilize in domestic auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $460 billion $9,174 billion 1990 in MAGICC emissions 2010 only part. -->2100 (-$11 billion) 2010 level (climate) 9/18/97 ID Scenario Model Modelers Target Timetable Trading Permit Revenue Burden BAU Paper Tons AEEI Ramp-up Time Path PDV (5%; GDPin 2010 Allocation Recycling Sharing Emissions or Corre- 2000-2050) (deviation Path sponding Foregone from BAU) Assumptio Reductions Consumption n SGM16 +10% of SGM, Battelle +10% 1990 stabilize in worldwide auction lump-sum no LDC IAT PT 1.0 in 2005 -->2050, $1 billion $9,184 billion 1990 in MAGICC emissions 2010 part. -->2100 (-$1 billion) 2010 level (climate) + MM1 BAU Markal DOE n/a n/a n/a n/a n/a n/a IAT n/a ~1.0 n/a -->2025 n/a $9,205 billion (BAU GDP) MM2 1990 in Markal DOE 1990 stabilize in domestic auction lump-sum no LDC IAT n/a ~1.0 n/a -->2025 n/a $9,137 billion 2010 emissions 2010 only part. (-$68 billion) level MM3 1995 in Markal DOE 1995 stabilize in domestic auction lump-sum no LDC IAT n/a ~1.0 n/a -->2025 n/a $9,152 billion 2010 emissions 2010 only part. (-$53 billion) level ~1.0 n/a -->2025 n/a $9,197 billion MM4 1990 in Markal DOE 1990 stabilize in domestic auction lump-sum no LDC TAT n/a 2020 emissions 2020 only part. (-$8 billion) level MM5 Peak in Markal DOE 2010 BAU stabilize in domestic auction lump-sum no LDC IAT n/a ~1.0 n/a -->2025 n/a $9,201 billion 2015 emissions in 2040 only part. (-$4 billion) 2015; 1990 level in 2040 9/18/97 ID Permit Permit Permit Conc. Conc. Year Conc. Change in Change in Emissions Year Intl. Trade of Intl. Trade of Date File (h:\jaldy\) Prices: Prices: Prices: (ppmv) in (ppmv) in (550 ppmv) Temp. (deg. Temp. (deg. Peak, Returns to Permits, U.S., Permits, U.S., Received 2010 2025 2050 2050 2100 Reaches 2x C) from 1990 C) from 1990 mmtce 1990 2010 2050 Run (Deviation (Deviation Pre-Ind. Level in 2050 in 2100 (year) (MMTCE), ($) (MMTCE), ($) from BAU) from BAU) (Deviation (Deviation (Deviation from BAU) from BAU) from BAU) SGM1 $0 $0 $0 502 711 2065 1.06 2.36 no peak: n/a n/a n/a 08/21/97 cea90_~1.xls 2245 (2050) SGM9 $91 $137 $238 481 645 2074 0.97 2.11 1637 never returns -180, -332, 08/21/97 cea90m_~1.xls (-21) (-66) (+9) (-0.09) (-0.25) (2005) (-$16.4 billion) (-$79.0 billion) SGM8 $175 $304 $924 481 645 2074 0.97 2.11 1637 2010 n/a n/a 08/21/97 cea90m_~1.xls (-21) (-66) (+9) (-0.09) (-0.25) (2005) SGM10 $33 LES $45 481 645 2074 0.97 2.11 no peak: never returns -332, -845, 08/21/97 cea90m_~1.xls (-21) (-66) (+9) (-0.09) (-0.25) 2060 (-$11.0 billion) (-$38.0 billion) (2050) SGM36 $23 $86 $150 486 656 2072 0.99 2.16 no peak: n/a -180 -364 09/10/97 case20~1.xls (-16) (-55) (+7) (-0.07) (-0.20) 1714 (-$4.14 billion) (-$54.6 billion) (2050) SGM39 $41 $84 $149 486 656 2072 0.99 2.16 no peak: never returns -202, -364, 09/10/97 case10~1.xls (-16) (-55) (+7) (-0.07) (-0.20) 1714 (-$8.3 billion) (-$54.2 billion) (2050) SGM3 $42 $84 $149 486 655 2072 0.99 2.15 no peak: never returns -202, -364, 08/21/97 cea90_~2.xls (-16) (-56) (+7) (-0.07) (-0.21) 1714 (-$8.5 billion) (-$54.2 billion) (2050) SGM18 $39 $83 $150 478 565 2091 0.96 1.82 no peak: never returns -209, -362, 08/27/97 case2.xls, (-24) (-146) (+26) (-0.10) (-0.54) 1712 (-$8.2 billion) (-$54.3 billion) 9/4/97 fax (2050) SGM21 $39 $83 $150 n/a n/a n/a n/a n/a no peak: never returns -209, -362, 08/27/97 case3.xls 1712 (-$8.2 billion) (-$54.3 billion) (2050) SGM17 $108 $188 $582 476 564 2092 0.95 1.82 1550 2010 n/a n/a 08/27/97 case2.xls, (2000) 9/4/97 fax (-26) (-147) (+27) (-0.11) (-0.54) SGM2 $110 $191 $582 485 658 2072 0.99 2.16 1637 2010 n/a n/a 08/21/97 cea90_~2.xls (-17) (-53) (+7) (-0.07) (-0.20) (2005) SGM38 $108 $188 $582 484 656 2073 0.98 2.15 1550 2010 n/a n/a 09/10/97 case10~1.xls (-18) (-55) (+8) (-0.08) (-0.21) (2000) SGM20 $108 $188 $582 n/a n/a n/a n/a n/a 1550 2010 n/a n/a 08/27/97 case3.xls (2000) SGM40 $16 $23 $33 486 656 2072 0.99 2.16 no peak: never returns -312, -751, 09/10/97 case10~1.xls (-16) (-55) (+7) (-0.07) (-0.20) 2101 (-$5.0 billion) (-$24.8 billion) (2050) 9/18/97 ID Permit Permit Permit Conc. Conc. Year Conc. Change in Change in Emissions Year Intl. Trade of Intl. Trade of Date File (h:\jaldy\) Prices: Prices: Prices: (ppmv) in (ppmv) in (550 ppmv) Temp. (deg. Temp. (deg. Peak, Returns to Permits, U.S., Permits, U.S., Received 2010 2025 2050 2050 2100 Reaches 2x C) from 1990 C) from 1990 mmtce 1990 2010 2050 Run (Deviation (Deviation Pre-Ind. Level in 2050 in 2100 (year) (MMTCE), ($) (MMTCE), ($) from BAU) from BAU) (Deviation (Deviation (Deviation from BAU) from BAU) from BAU) SGM4 $16 $23 $32 486 655 2072 0.99 2.15 no peak: never returns -313, -752, 08/21/97 cea90_~2.xls (-16) (-56) (+7) (-0.07) (-0.21) 2102 (-$5.0 billion) (-$24.1 billion) (2050) SGM22 $15 $23 $0 n/a n/a n/a n/a n/a no peak: never returns n/a n/a 08/27/97 case3.xls 2239 (2050) SGM19 $15 $23 $110 478 565 2091 0.96 1.82 1883 never returns -316, -500, 08/27/97 case2.xls, (-24) (-146) (+26) (-0.10) (-0.54) (2040) (-$4.7 billion) (-$55.0 billion) 9/4/97 fax SGM35 $71 $192 $582 484 656 2073 0.98 2.15 1550 2020 n/a n/a 09/10/97 case20~1.xls (-18) (-55) (+8) (-0.08) (-0.21) (2000) SGM37 $9 $24 $33 486 656 2072 0.99 2.16 no peak: n/a -246 -751 09/10/97 case20~1.xls (-16) (-55) (+7) (-0.07) (-0.20) 2101 (-$2.21 billion) (-$24.78 (2050) billion) SGM6 $74 $119 $202 483 648 2074 0.98 2.12 1637 never returns 36, -122, 08/21/97 cea95_~1.xls (-19) (-63) (+9) (-0.08) (-0.24) (2005) (+$2.7 billion) (-$24.6 billion) SGM5 $62 $131 $317 483 648 2074 0.98 2.12 1637 never returns n/a n/a 08/21/97 cea95_~1.xls (-19) (-63) (+9) (-0.08) (-0.24) (2005) SGM7 $27 $32 $41 483 648 2074 0.98 2.12 no peak: never returns -131, -593, 08/21/97 cea95_~1.xls (-19) (-63) (+9) (-0.08) (-0.24) 2073 (-$3.5 billion) (-$24.3 billion) (2050) SGM33 $57 $120 $203 482 648 2074 0.97 2.12 no peak: n/a 20 -119, 09/10/97 case20~2.xls (-20) (-63) (+9) (-0.09) (-0.24) 1599 ($1.14 billion) (-$24.16 (2050) billion) SGM32 $51 $131 $318 482 648 2074 0.97 2.12 1550 n/a n/a n/a 09/10/97 case20~2.xls (-20) (-63) (+9) (-0.09) (-0.24) (2000) SGM34 $21 $33 $41 482 648 2074 0.97 2.12 no peak: n/a -122 -592 09/10/97 case20~2.xls (-20) (-63) (+9) (-0.09) (-0.24) 2072 (-$2.56 billion) (-$24.27 (2050) billion) SGM12 $0 $47 $155 491 n/a n/a 1.02 n/a 1807 never returns 0, (n/a) -367, 08/21/97 cea90_~1.xls (-11) (-0.04) (2015) (-$56.9 billion) SGM30 $11 $56 $153 489 658 2071 1.00 2.17 1755 never returns -5, -367, 08/27/97 case6.xls, (-13) (-53) (+6) (-0.06) (-0.19) (2015) (-$0.06 billion) (-$56.2 billion) 9/4/97 fax SGM24 $11 $56 $153 481 568 2089 0.98 1.84 1755 never returns -5, -367, 08/27/97 case4.xls, (-21) (-143) (+24) (-0.08) (-0.52) (2015) (-$0.06 billion) (-$56.2 billion) 9/4/97 fax 9/18/97 ID Permit Permit Permit Conc. Conc. Year Conc. Change in Change in Emissions Year Intl. Trade of Intl. Trade of Date File (h:\jaldy\) Prices: Prices: Prices: (ppmv) in (ppmv) in (550 ppmv) Temp. (deg. Temp. (deg. Peak, Returns to Permits, U.S., Permits, U.S., Received 2010 2025 2050 2050 2100 Reaches 2x C) from 1990 C) from 1990 mmtce 1990 2010 2050 Run (Deviation (Deviation Pre-Ind. Level in 2050 in 2100 (year) (MMTCE), ($) (MMTCE), ($) from BAU) from BAU) (Deviation (Deviation (Deviation from BAU) from BAU) from BAU) SGM27 $11 $56 $153 n/a n/a n/a n/a n/a 1755 never returns -5, -367, 08/27/97 case5.xls (2015) (-$0.06 billion) (-$56.2 billion) SGM11 $0 $84 $559 489 n/a n/a 1.01 n/a 1807 2040 n/a n/a 08/21/97 cea90_~1.xls (-13) (-0.05) (2015) SGM29 $12 $99 $563 489 658 2071 1.00 2.17 1729 2040 n/a n/a 08/27/97 case6.xls, (-13) (-53) (+6) (-0.06) (-0.19) (2015) 9/4/97 fax SGM26 $12 $99 $563 n/a n/a n/a n/a n/a 1729 2040 n/a n/a 08/27/97 case5.xls (2015) SGM23 $12 $99 $563 481 568 2089 0.98 1.84 1729 2040 n/a n/a 08/27/97 case4.xls, (-21) (-143) (+24) (-0.08) (-0.52) (2015) 9/4/97 fax SGM25 $4 $16 $111 481 568 2089 0.98 1.84 1893 never returns -30, -500, 08/27/97 case4.xls, (-21) (-143) (+24) (-0.08) (-0.52) (2030) (-$0.1 billion) (-$55.5 billion) 9/4/97 fax SGM31 $4 $16 $33 489 658 2071 1.00 2.17 no peak: never returns -30, -751, 08/27/97 case6.xls, (-13) (-53) (+6) (-0.06) (-0.19) 2101 (-$0.12 billion) (-$24.8 billion) 9/4/97 fax (2050) SGM13 $0 $14 $33 491 n/a n/a 1.02 n/a no peak: never returns 0, (n/a) -752, 08/21/97 cea90_~1.xls (-11) (-0.04) 2102 (-$24.8 billion) (2050) SGM28 $4 $16 $0 n/a n/a n/a n/a n/a no peak: never returns n/a n/a 08/27/97 case5.xls 2241 (2050) SGM15 $17 $56 $113 489 662 2071 1.00 2.18 no peak: never returns -172, -339, 08/27/97 casel.xls (-13) (-49) (+6) (-0.06) (-0.18) 1824 (-$2.9 billion) (-$38.3 billion) (2050) SGM14 $60 $128 $310 488 664 2071 0.99 2.18 1550 never returns n/a n/a 08/27/97 case1.xls (-14) (-47) (+6) (-0.07) (-0.18) (2000) 9/18/97 ID Permit Permit Permit Conc. Conc. Year Conc. Change in Change in Emissions Year Intl. Trade of Intl. Trade of Date File (h:\jaldy\) Prices: Prices: Prices: (ppmv) in (ppmv) in (550 ppmv) Temp. (deg. Temp. (deg. Peak, Returns to Permits, U.S., Permits, U.S., Received 2010 2025 2050 2050 2100 Reaches 2x C) from 1990 C) from 1990 mmtce 1990 2010 2050 Run (Deviation (Deviation Pre-Ind. Level in 2050 in 2100 (year) (MMTCE), ($) (MMTCE), ($) from BAU) from BAU) (Deviation (Deviation (Deviation from BAU) from . BAU) from BAU) SGM16 $7 $16 $25 489 662 2071 1.00 2.18 no peak: never returns -219, -643, 08/27/97 casel.xls (-13) (-49) (+6) (-0.06) (-0.18) 2128 (-$1.5 billion) (-$16.1 billion) (2050) MM1 $0 $0 n/a n/a n/a n/a n/a n/a no peak: n/a n/a n/a 08/26/97 8_26runl.wk4 2066 (2025) MM2 $148 $192 n/a n/a n/a n/a n/a n/a 1586 2010 n/a n/a 08/26/97 8_26runl.wk4 (2005) MM3 $136 $146 n/a n/a n/a n/a n/a n/a 1600 n/a n/a n/a 08/26/97 8_26runl.wk4 (2005) MM4 $0 $198 n/a n/a n/a n/a n/a n/a 1749 2020 n/a n/a 08/26/97 8_26runl.wk4 (2010) MM5 $0 $99 n/a n/a n/a n/a n/a n/a 1767 2040* n/a n/a 08/26/97 8_26runl.wk4 (2015) William D. Nordhaus Yale University Department of Economics Tel: 203-432-3587 28 Hillhouse Avenue Fax 203-432-5779 New Haven, Connecticut 06511-8268 June 20, 1997 see Figure 9 Dr. Jay Shogren Council of Economic Advisers Dear Jay, This is a refined version of the earlier paper. The RICE-model simulations have improved and do not look good for Rio-type schemes either from an economic or political point of view. book: when things Sincerely, bite back Rin CLIMATE ALLOWANCES PROTOCOL (CAP): COMPARISON OF ALTERNATIVE GLOBAL TRADABLE EMISSIONS REGIMES William D. Nordhaus June 17, 1997 Preliminary Abstract Recently, the U.S. government proposed an international emissions-trading protocol to cope with the threat of global warming due to emissions from greenhouse gases (GHGs). While such a proposal is superior to earlier approaches (such as the country-specific emissions targets of the Rio Treaty or the Berlin protocol), it leaves open a number of crucial questions. These include the appropriate objectives, the distribution of emissions budgets among countries, a mechanism to discover countries' perceived benefits from such an arrangement, as well as a set of incentive and enforcement procedures. This paper develops a Climate Allowances Protocol (CAP) that addresses many of the outstanding issues. It is based on the view that the system should balance costs against benefits, be economically efficient, and respect the realpolitikal fact that sovereign nations must find it in their self-interest. The CAP plan integrates climate policy with an underlying cost-benefit approach that maximizes global incomes and allows countries to arrive at a level of the public good that reflects an aggregation of individual country preferences. Simulations of the efficiency and realism of the CAP plan using an updated version of the RICE model indicate that it has major economic advantages, leads to benefits for most participants, and has equivalent environmental benefits as compared to current international proposals. -1- CLIMATE ALLOWANCES PROTOCOL (CAP): COMPARISON OF ALTERNATIVE GLOBAL TRADABLE EMISSIONS REGIMES William D. Nordhaus¹ June 17, 1997 Preliminary 1. Background and Motivation Recently, the U.S. government proposed an international emissions-trading protocol to cope with the threat of global warming due to CO 2 emissions from combustion of fossil fuels and other greenhouse gases (GHGs). 2 While a market mechanism is superior to earlier approaches (such as country-specific emissions targets), it leaves open a number of crucial questions. These include the appropriate ultimate concentrations and emissions targets, the distribution of allowable emissions among countries, a mechanism to discover countries' perceived benefits from such an arrangement, as well as the issues of incentive and enforcement procedures. To date, there has been virtually no attention to the economic aspects of the design of such a protocol, and surprisingly little attention to making the design consistent with sound economic and political principles. This paper describes a regime (the "Climate Allowances Protocol" or CAP) that takes into account current economic theory of public goods and compares this approach with current proposals that are being discussed in government circles. The basic philosophy here is that the system should balance costs against benefits, be economically efficient, and respect the fact that the agreement must be among sovereign nations and must be in the self-interest of participating nations. Global warming poses especially nasty obstacles to reaching efficient and effective policies. It is a public good over space and time in the sense that emissions in any place affect climate everywhere and for centuries in the future. There is no obvious technological fix. The scientific and economic uncertainties are sufficiently large that those who would be harmed by policies can easily generate a smokescreen to confuse people by arguing that the 1 The author is A. Whitney Griswold Professor of Economics and on the staff of the Cowles Foundation, 28 Hillhouse Avenue, Yale University, New Haven, CT 06520, USA. 2 See U.S. Department of State, U.S. Draft Protocol and Fact Sheet on Climate Change Proposal, January 17, 1997. -2- scientific case is weak or trumped up. And the economic stakes are enormous, with the present value of costs and future damages of many proposals likely to run in the hundreds of billions of dollars annually. These features are exacerbated by the fact that there is no obvious anchor or "focal" point for policy. In this respect, global warming differs from other policies that have focal strategies, such as trade regimes (zero tariffs and trade barriers) or nuclear non-proliferation agreements (no weapons or weapons-grade material). Without focal policies, agreement is more difficult because reasoned disagreement is so easily sustained. While this discussion is aimed at the threat of greenhouse warming, I think of it as a prototype for any important global economic public good. A global economic public good in one for which there are huge numbers of economic agents in a large numbers of countries and where the costs and benefits of action do not indicate any obvious focal policy or technological fix. While global warming might or might not turn out to be a critical issue for humanity or natural ecosystems, it is useful to think through how we might design a global economic regime when and if a crucial global economic public good appears on the radar screen. If you think global warming is a critical issue, this should interest you; if not, then imagine a global economic public good that is critical and apply these principles there. This sketch will begin with a summary of the mechanisms and then proceed to see how it addresses the major design hurdles. It is meant only to sketch the major issues because a full treatment would be longer than the tolerance of most mortals.³ 2. The Thorny Issues of Regime Design Current international proposals for limiting greenhouse gases (such as those set out in the Rio Treaty or in the current U.S. proposal) are deeply flawed. They are based on arbitrary historical benchmarks for emissions limitations and make no attempt to link the path of emissions to any economic or environmental objective. They provide no incentives for countries to join or to follow the rules once they have joined. It is not surprising that every serious economic analysis of these proposals finds them highly inefficient. 3 This discussion will contain only a few references from the vast literature on the subject. An excellent overview with a bibliography to most of the literature is contained in B.S. Fisher, S. Barrett, P. Bohm, M. Kuroda, J.K.E. Mubazi, A. Shah, and R.N. Stavins, "An Economic Assessment of Policy Instruments for Combatting Climate Change," Chapter 11 in James P. Bruce, Hoesung Lee, and Erik F. Haites, Climate Change 1995: Economic and Social Dimensions of Climate Change, Cambridge University Press, Cambridge, U.K., 1996, pp 397-439. Many of the legal principles are laid out in Stewart??? -3- What first principles would be look to in designing a regime that is economically efficient and politically viable? The most important characteristics are the following: The regime should be moderately efficient from an economic and administrative point of view. There must be positive incentives for countries to participate in the regime and to fulfill their obligations. The regime must contain a decision process which will induce countries to agree to efficient levels of greenhouse gas reduction. I believe that the proposal discussed below satisfies the first three criteria reasonably well. The next three points are ones that are troubling and have no obvious solutions: The regime must recognize that countries with inept or corrupt governments may subvert the process or use it for personal gain. The regime must not injure other international agreements, particularly international trade agreements. The decision process should be simple and familiar to countries. 3. The Basic Mechanism The alternative regime considered here - which is called the Climate Allowance Protocol (CAP) - would embed the price and quantity targets within a framework that considers both environmental and economic objectives and sets the policies to maximize net benefits. CAP proceeds by having a set of tradable emissions allowances or permits with carbon price ceilings and floors. The major innovation is that the emissions limits and carbon prices are derived from a dynamic cost-benefit analysis that incorporates all costs and benefits of slowing climate change. The costs are estimated from engineering and economic models, and will be tested in each accounting period. The benefits will be derived from a public goods mechanism of participating countries. Participant countries will be allocated a number of emissions permits for each budget period. A major and fundamental difference between the CAP and current proposals is that the emission baselines for individual countries are rolling targets rather than ones set on the -4- basis of an arbitrary historical benchmark. In addition, there are expectations that countries will begin to participate as their economies develop. In CAP, countries will be expected to make small emissions reductions when their per capita incomes reach an "entry-level income" and to participate fully when their incomes reach "full-participation income." In each budget period, the number of permits will be allocated according to a formula that reflects each country's uncontrolled emissions and its ability to pay. Because carbon prices are equalized across participating countries, there will be no need for tariffs or border tax adjustments among participants. The incentive to participate - and sanctions against violations - will come from import duties on the carbon content of imports of non- participants into participating countries. Unlike current proposals, there is no need for financial transfers among countries, which raise severe political complications. The balance of the paper describes these points in detail. 4. The Trading Rules The basic approach of the Climate Allowance Protocol is to have an agreement among the participating countries to limit the total annual emissions of greenhouse gases. The overall limitation will be allocated to individual countries, who will receive their basic allowances of greenhouse gases. Nations will be allowed to buy, sell, barter, and borrow other countries' allowances as long as all the rules are obeyed. The fundamental rule is that a country's emissions shall not exceed the sum of its own allowances plus net purchases of allowances from other countries. Emissims:= Eij e, & o CAP would initially include "covered fuels," which comprise the consumption of fossil fuels. For accounting purposes, consumption will be equal to domestic production of fossil fuels plus net imports. For example, for the U.S., consumption would include production of crude oil, natural gas, coal, as well as net imports of those fuels and net imports of petroleum products such as gasoline. Consumption would exclude derivative products such as petrochemicals, electricity, chemicals, as well as the fuels embodied in manufactured products. In the CAP, because emissions limits are pegged to uncontrolled ? emissions there are no significant distributional or efficiency impacts of the exact definition CRUDE oil Production Natural gas & gasoline - imports COAL Imports emissime limits pegged to uncontrolled emissions -5- of covered fuels. 4 There is a complication for imports from non-participants, which is discussed later in the paper. Each nation will be allowed to administer its emissions limitations in its own manner. In a market economy, it would be natural to require that each producer or importer of covered fuels purchase the requisite emissions allowances. For reasons discussed below, it is recommended that governments auction or sell their allowances. We estimate that the cost of X the allowances would be shifted forward to consumers except for fuels that are highly price- inelastic in supply. 5. Balancing Costs and Benefits in the Emissions Trading Regime The fundamental philosophy of the CAP is that global emissions targets are chosen to balance costs and benefits of emissions reductions. This approach is radically different from current approaches in international negotiations (such as those under the Rio Treaty), which have targets and timetables for emissions limitations with no rationale for the targets and screnti timetables. The Rio emissions limits are ones that occurred during a particular historical economic period in a particular country; they are easy to understand and explain. But they make no ratural sense; there is no link between the targets and timetables, on the one hand, and the economic or environmental benefits. Need to L ink The target d timetribles to econ. benefits. The proposal here is to link the targets explicitly to economic and environmental benefits through the use of dynamic cost-benefit analysis. The state of the art in cost-benefit analysis of climate change are "integrated assessment" (IA) models, of which a wide variety are currently available and in use.⁵ Most of the inputs into these models are of a scientific and technocratic nature. Such inputs include parameters of the carbon cycle, engineering and scientific data on global warming potential and carbon content of different fuels, and 4 The general principle is to define covered fuels in a manner that monitoring and enforcement costs are minimized. For that purpose, covered fuels are production of fossil fuels plus net imports of fuels and direct products. The only subtlety here is that consumption includes net imports, which will require some careful thought to ensure consistency and efficiency. Under the principles for allocation of allowances used here, the definition is of no importance for efficiency and is only of second-order importance for distribution. The definition would have great importance under current international proposals that have a fixed as opposed to a rolling base. 5 An excellent overview of Integrated Assessment models by lead author John Weyant and collaborators is contained in Chapter 10 of James P. Bruce, Hoesung Less, and Erik F. Haites, eds., Climate Change 1995 - Economic and Social Dimensions of Climate Change, Cambridge University Press, 1996. -6- 1. optimizen M b/c 2. optional emissims over 5 year. 3. climate models. Two sets of non-technocratic data are, however, crucial for the results: the costs of emissions reduction and the damages from (or willingness-to-pay to avoid) climate change. Derivation of these will be described below. The full output of the dynamic cost benefit analysis will be an "indicative plan" for global GHG emissions and carbon taxes. This indicative plan will be the basis of the overall limits on emissions. For each (accounting period (say five years), countries will be allocated emissions allowances whose sum equals the optimized emissions from the IA models. The CAP mechanism relies on allocating emissions rather than carbon taxes because there is no mechanism for allocating carbon taxes. But carbon taxes or prices are a central part of the plan and will be introduced by having price ceilings and floors. Should emissions limits within a budget period be tighter than expected because of model mis-estimates, countries may buy emissions allowances at a ceiling price that is a premium (say, 150 percent) of the optimized carbon tax. Too-low prices will be ignored for the moment. Deviations of prices or quantities from the levels projected in the IA planning models for the budget period will automatically be corrected by model revisions for the next budget period. This automatically incorporates a procedure by which the latest scientific and economic information and revised national priorities will be incorporated into the prices and quantities each period. ADJUSTABLE TARgets given New intromation M costs d 6. Costs of Emissions Control benefits In our summary, we stated that the cost side of the cost-benefit analysis would be automatically revealed in the emissions-trading market. We explain that point in this section. The revelation is seen most clearly for a private market economy. Suppose that the price of an emissions permit is = $10 per ton of carbon. This means that firms can produce output using covered fuels at a marginal cost of P+T or substitute inputs at a cost of R. This implies that the marginal cost of reducing emissions is MC = R - P The revelation of marginal cost in illustrated in Figure 1. The curves MC and MC² are the marginal cost curves for emissions reductions in regions 1 and 2. The efficient global cost-of-reduction curve is shown as MCᵀ. The vertical line at E' is the sum of the emissions limitations of the regions, = How ave R=P+2 E' dE² E' MC CF = P+2 chopsen? MC oner = R R-P=2 Firms in each regions will purchase or sell permits to the point where their MC of Marginal cost, carbon tax emissions reduction equals the È' E² ET permit price. The efficient allocation is shown in Figure 1 at a permit price of τ*. At that MC' MC² MC¹ price, region 1 buys permits shown by the segment A, while region 2 sells segment B. The balancing condition is B + A = A 0. B The key point to note is 7 that (for cost minimizing E' E² E 2 È ET entities), the permit price will be equal to the marginal cost of Figure 1. Emissions cost curves in regions 1 and 2 and emissions reductions for each equilibrium price of permits region. We will thus be able to observe the marginal cost curve for emissions reductions as individual entities make their decisions. NGA FREIDA 7. Public Goods Mechanism In contrast to cost of controls, the value of emission reductions (or averted damages of climate change) cannot be determined by a market mechanism. Most of the damages will be felt far in the future, and indeed even the "experts" have only the haziest notions about what those costs may be and who will bear them. It is therefore necessary to design a process whereby reliable estimates of different countries' willingnesses to pay can be derived. Slowing harmful climate change is a global public good. Individual people, countries, and generations make but a small contribution to slowing climate change, and this leads to the possibility of free-riding, posturing, temporizing, and insincere statements of position. On top of this intrinsic difficulty is the further complication that environmental policy is largely in the hands of environmentalists and large companies, neither of which adequately represents the interest of consumers. A final difficulty is that global warming involves countries with highly divergent incomes and priorities. The major difficulty in any plan, therefore, is to design a public goods mechanism that can find a tolerably good solution to this very nasty public goods problem. -8- The mechanism proposed here responds to these problems by containing two stages. The first stage is the constitutional stage. In this part, countries agree on the basic framework of the process. This stage involves the choice of international instrument, the accounting framework, the voting procedure, setting sanctions for non- participation or infractions, and the legal ratification and amendment processes. Given the long time horizon of the global warming problem, this stage effectively operates behind a Rawlsean veil of ignorance because costs and particularly benefits lie far in the future compared to most current investments and elections. Given the nature of the public good, this stage is largely distributional and could easily lead to protracted impasses. The second stage contains the decision-administrative processes. Given the constitution of the global warming agreement, heterogenous country preferences will have to be aggregated into a single global level of emissions. Among the alternative processes we suggest that a voting mechanism is likely to be the most fruitful. Before discussing the constitutional and decision-administrative phases, some background comments are necessary. The mechanism proposed here treats the massive public goods problem by having nations serve as trustees of future generations and by separating the costs today from the benefits in the future. The central idea is that countries will vote a given "value" of reducing climate change (or the willingness to pay [WTP] to reduce slow climate change). When the different values are aggregated, this will imply an efficient level of emissions control and an associated carbon emissions price or carbon tax today. None of these numbers will be a surprise to the countries who are voting on the decision -- rather the final outcome will be the result of negotiations, persuasion, recalibration, models runs, bargaining, and further negotiations until an agreed level of control and carbon price is agreed upon. It should be emphasized that the design of the public goods mechanism is both the core of the proposal and its most problematic feature. While countries routinely use voting mechanisms to decide upon national public goods, there are a very few supernational examples of functional public goods mechanism. Those used by the European Union may be the closest in nature to the proposal here. 8. The Constitutional Stage Before voting, countries need to know how they will be "taxed" to pay for the public good of slowing climate change. The mechanism outlined here presumes that nations are -9- likely to vote resources to emissions reductions primarily out of concern for the future rather than because they believe that they are likely to be differentially affected. Nations which have high willingness to pay and low discount rates will select higher control rates and carbon emission prices than those who are primarily concerned with today's economic and political issues. The payment mechanism will start with a zero base of uncontrolled emissions. This initial condition reflects the realistic fact of life that the status quo of doing nothing has a privileged position under custom and international law. Starting from the status quo, countries then pay according to both horizontal and vertical equity principles. The horizontal equity principle is that countries with the same per capita incomes should contribute proportionally to their current emissions. Assuming that price elasticities of supply and demand are equal across countries, this implies that each horizontally equivalent country shall receive permits that are a fixed fraction of baseline or uncontrolled emissions; equivalently, this means that the expected emissions reduction rate is equal for all horizontally equivalent country. If the elasticities are equal, this implies that the cost of participating in and complying with such a mechanism will be an equal fraction of national income for all countries that have the same level of emissions. The vertical equity principle would be that poor countries should be required to pay less than rich countries. The formula that is explored below proposes a floor of $5000 per capita (below which a nation would have no expectations of emissions reduction) and a ceiling of $15,000 per person (above which a nation would be expected to have full participation). The participation rate would then vary linearly over that range. (I consider alternative progressivity schedules below.) Formally, nation i would in accounted period t receive a number of emissions permits equal to Eᵢₜ = E₀,ᵢ,t [(1-f(yᵢ,)µ,*] Here E₀,ᵢ,ₜ is the baseline or uncontrolled level of emissions in period t, f(yᵢ) is a fraction running from 0 to 1 as a function of the country's per capita income (yᵢ), and µ,* is the agreed target level of global emissions reduction for full participants in period t. The actual level of global emissions reduction (µ,ᵃᶜᵗ) will be less than the maximum level (µ,*) depending upon the degree of participation, the progressivity of the vertical equity schedule f, and the degree of compliance. The total Σ 0,i,t,f(yᵢ) is the amount of emissions that is subject to control and is called "reducible emissions." The functioning of the mechanism is sketched in Figure 2. Estimates of the marginal costs of emissions are based on engineering and economic models for the first period. In -10- seems subsequent periods, more reliable estimates will be Marginal cost, carbon tax obtained as countries begin MCᵀ reducing emissions and the price of permits rises above zero. The marginal benefit function is derived from the voting mechanism described below. These two fundamental building * blocks are then used to MBᵀ generate the associated allowable emissions [(1- E* Allowable emissions µᵃᶜᵗ)E₀] and carbon permit Emissions reduction E₀ price (t*) as shown in Figure 2. Figure 2. Derivation of Allowable Emissions In each accounting period, MCᵀ is estimated global marginal With no cost of emissions reductions while MBᵀ is the estimated uncertainty, we could in marginal benefit from emissions reduction derived from public principle use either the tax goods mechanism. This produces indicative carbon tax (t*) or emissions limitations. and allowable emissions E* = (1-u)E₀. Emissions reductions In fact, the tax approach is are then reallocated among countries according to region- unworkable because it specific MC curves and trade in permits as shown in Figure 1. cannot be accurately verified against a background of other policies (such as fuel taxes and coal subsidies). We therefore use the emissions limitation approach as the most easily administered. Given the uncertainty about the cost function, it is not sensible to ignore prices, however. To avoid costly mistakes, it is suggested that a cap on carbon prices be set at some premium (say 50 percent) over the estimated efficient price. (It may be possible to determine the optimal premium from IA models.) The revenues from the sale of extra permits at the premium price can be used as determined by the participating countries. For the case of prices that fall below the estimated price, it is suggested that countries be allowed to "bank" the excess permits for future use. We leave open the question of whether it would be useful to have a fund for stabilization purposes. -11- 9. The Voting Mechanism The second stage of the process is the decision-administrative process in which the revelation of the value of the public good is the essential issue. From an analytical perspective, there are many proposals for mechanisms to facilitate revelation of preferences for public goods. Most share the feature that they are "one-dimensional" in the sense that agents vote for a level of public goods with a payment mechanism in place. The major desideratum of an effective public goods mechanism is that it be incentive compatible. An incentive compatible mechanism is one in which agents have an incentive to reveal their true preferences. There are a large number of potential ways that countries can decide upon a global emissions-reduction target. 6 (a) One approach is simply to allow countries to decide on their own. In the case of a Nash equilibrium where countries pursue only their own self-interest, calculations indicate that the degree of emissions reduction would be minuscule. (b) The Lindahl equilibrium asks countries to reveal their own individual prices and then sets the NOT global rate of emissions reduction to maximize global welfare. This is defective because TRUE countries have strong incentives to misstate their own preferences. (c) One approach is simply allow countries to bargain until they reach agreement. Under the so-called Coase theorem, this process would reach an efficient outcome. (The transactions costs here are trivial relative to the economic costs and benefits.) While bargaining might reach a solution, it seems more likely to get bogged down and to be subject to serious free riding by non- participants. (d) A theoretically attractive approach is to use the Clarke-Groves mechanism. In this approach, countries bid the value of different global warming strategies. They are then taxed on the basis of their bids, with individual taxes calculated as the contribution of each country to the net cost imposed on other countries. This mechanism has attractive theoretical properties but is extremely complex, is far to intricate to explain to busy politicians, and has never to my knowledge been implemented. The approach proposed here is to use a voting mechanism. This has the advantage that it is incentive compatible, is likely to come close to the efficient outcome (as long as tastes are not too skewed), and is familiar to most people. The major disadvantage compared to efficient public-goods mechanisms or bargaining is that it can impose significant costs on those whose preferences deviate significantly from the mainstream. The basic idea is the following: 6 I omit what seem fruitless or highly inefficient approaches like moral suasion and mandatory technical standards. While these are highly popular among politicians and can sometimes be useful supplements, they cannot form the core of an efficient and effective global public goods mechanism. -12- Voting Mechanism. The rate of emissions reduction is determined on the basis of the globally aggregated willingness to pay for reducing climate change. The willingness to pay is aggregated from the individual willingnesses to pay of different countries as expressed by their votes. The individual willingnesses are then averaged across participants on the basis of their reducible emissions. In other words, the targeted emissions reduction is decided by voting of participants in the plan. Each country expresses its willingness to pay (call it parameter wᵢ). The individual countries willingnesses are then aggregated into a global willingness (Ω) by averaging across the participants. Using the globally averaged willingness, then, we can estimate the optimal emissions reduction and carbon tax or permit price. To make this more concrete, we would define the willingness to pay parameter as the annual willingness to pay to avoid the damages accompanying an equilibrium doubling of CO₂ (or alternatively of a 3 degrees C warming). For example, a country might express a willingness to pay 2 percent of GDP to prevent the equilibrium climate change associated with a doubling of atmospheric CO2. To obtain the economically correct emissions reduction, we would use the global average willingness to pay (Ω = Ew 2ᵢ) which is obtained by averaging the individual country willingness to pay (wᵢ) by their shares of a world income (xᵢ). Unfortunately, this presents serious incentive problems. Weighting WTP by future incomes or total emissions (which is the appropriate measure of damage) would allow rapidly growing countries or countries with low emission reductions fractions (f) to free ride on current emitters. To avoid free riding, we need to work with a variant of global WTP that is incentive compatible. To do this, we weight each nation's expressed WTP by its share of reducible emissions. This mechanism will give an unbiased estimate of current WTP if WTP is uncorrelated with f(yⱼ). Weighted median voting is simple, robust to strategic distortions, and prevents free riding: Weighted medium vot; Voting Mechanism. The voting mechanism in the plan would be to set the global willingness to pay (Ω) equal to the median WTP. In this calculation, each unit of reducible emissions would have an equal vote. This mechanism is incentive compatible because there is no advantage for any country with single-peaked preferences⁷ to vote anything other than its most preferred policy. To do 7 All studies of the damage from climate change indicate that countries have single-peaked preferences with respect to the willingness-to-pay parameter. -13- otherwise would either have no effect if it did not change the outcome or would lower its economic welfare if it did change the outcome. Huh? 10. Calculation of Baseline Emissions Baseline emissions play a critical role in the mechanism because determine the economic interests of countries, affect the equity of the system, and have a major impact upon incentive compatibility. One of the major flaws in current proposals (such as the Rio Treaty or the 1996 U.S. emissions-trading protocol) is that they rely on historical emissions to determine future allowances. This leads to growing inequities among countries and provides a major roadblock to reaching an agreement. Under current proposals, countries have emissions limits that are determined by their 1990 emissions levels. This penalizes efficient countries (like Sweden) or rapidly growing countries (like Korea). It also gives a premium to slow-growing countries (like Britain) while historically inefficient countries (like Russia or Germany after unification) are rewarded by having too high a base. This point can be illustrated as follows. We estimate that industrial CO₂ emissions for the U.S. in 2000-2009 will be about 1.64 billion tons of carbon per year. This represents a growth of 18 percent over the prior decade and an increase of about 20 percent over 1990 emissions. Under current proposals, the U.S. would continue to live with its 1990 base in the indefinite future. By contrast, Russia and Germany had emissions in 1994 of about 20 percent less than 1990 levels; they would receive large windfalls if the 1990 base were continued in the future. One of the major distinguishing features of the CAP is the calculation of each country's base emissions. Emissions base: Each country's emissions base would be equal to its uncontrolled emissions in each budget period. This base would roll forward with the change in each country's growth in output and population. Under CAP, a country's base for reductions would be its uncontrolled level of emissions. This base would then roll forward with each budget period. Countries that experience rapid growth in their economies would not thereby be penalized, nor would countries receive windfalls if they were historically inefficient. There definitely would be penalties from higher emissions, but penalty would be proportional to the level of emissions rather than some irrelevant historical experience. Aside from better equity properties, the CAP approach of a rolling base also improves the possibilities of reaching an agreement. As noted above, the CAP approach has the -14- feature that it would lead to a unanimous decision if countries have the same demand and supply elasticities and the same damages from climate change (even if their growth rates differ). The historical base has the unfortunate feature that the decision will depend upon the growth rate; indeed, some countries (like Germany or Russia) might vote for very large emissions reductions because they have such large emissions bases relative to their actual emissions that this would drive up the price of their emissions allowances. Other countries like the U.S. or Japan with growing emissions would find themselves with a severe relative disadvantage. Indeed, they might even find that their optimum would be for negative emissions reductions given their small base. One problem with the rolling base is the possibility of countries having an incentive to "ratchet" up their emissions in one period to increase the base in the next period. The "ratchet effect" is a phenomenon observed in planning systems wherein individual agents have incentives to increase spending (or in this case emissions) to increase their base for the next planning period. Two points are important here. First, ratchet effects at the individual or explicit firm level are minuscule because these agents have such a tiny impact upon national collusion emissions levels. Second, while the rolling base definitely attenuates the incentive to reduce emissions at the country level, it does not give the wrong incentives as long as the real price of allowances is not rising too rapidly. For constant emissions reduction rates, the ratchet effect will be neutral when the carbon permit price is rising at the real interest rate. In the more likely case where the carbon permit price is rising more slowly than the real interest rate, the value to a country of an additional unit of emissions is negative. Again, note that the ratchet effect applies only to countries, who receive permits, and not to individuals, who must buy them. 8 11. Participation and Sanctions Whatever the moral imperative to protect the planet, it seems likely that the present costs will outweigh future benefits for most countries unless there is a cooperative arrangement with sanctions on disinclined and wayward countries. One indication of the tendency for free-riding is the estimate of the Nash-equilibrium. If countries take into account only their own national interests, the control rates and carbon prices are likely to be only a small fraction of the cooperative prices. Countries are unlikely to benefit from global 8 The present value of a unit of emissions in period t is -t, + ]. In this expression, t, is the price of permits in accounting period t, µ₁ is the emissions control rate, r is the real interest rate, and Θ is the length of the accounting period. This value is negative as long as the permit price is rising less than the real interest rate and as long as the control rate is not falling. These conditions are met in virtually all simulations. -15- warming policies for at least half a century.⁹ To make joining such an agreement for individual countries worthwhile, therefore, not only must it impose reasonable, efficient, and equitable burdens, but it must also impose serious and proportional sanctions on the wayward and those who will not participate. The most workable sanction would be carbon import duties levied on non- participating and non-complying countries. There would be a number of different sanction designs, but the simplest would be to have participating countries impose duties equal to the current price of emissions permits times the carbon content of the imports of non-complying countries into participating countries. For example, suppose that the U.S. were a participant which persistently ignored its targets, which were a 10 percent reduction from a base emissions of 1640 million tons of carbon. If the carbon emissions fee were $10 per ton, the U.S. would face an import duty of that amount on its exports into participating countries. If the U.S. exported one-fifth of its carbon output, then this would amount to duties of $10 X 0.2 X 1640 = $3.28 billion annually. The estimated annual cost of compliance is, by my estimate, $0.85 billion, so it would definitely pay for the U.S. to comply rather than face the import duties. 9 This point is clearly shown in W. Nordhaus and Z. Yang, "A Regional Dynamic General Equilibrium Model of Optimal Climate-Change Policy," American Economic Review, September 1996. -16- The tradeoff is shown in Figure 3 for a country Marginal cost, carbon tax considering whether or not to join the plan. Compliance costs MC are area A under the marginal cost curve. Noncompliance would cost the country the current carbon price times the carbon content of exports, shown as area B in Figure 3. Carbon Exports For small open economies, relatively low emissions reductions, and a large group of B participating countries, the A incentives to participate are Required emissions reduction Allowable emissions E₀ likely to be quite strong. Numerical calculations indicate Figure 3. Country can comply by incurring costs of that if the plan gets off the emissions reductions of A. Non-complying countries face ground by including more than import duties on exports x carbon tax shown as area B. half of countries, and if it is not overly ambitious, the incentive to participate would be powerful except for very closed economies. 12. Corruption and Standards for Country Behavior A mechanism for emissions trading contains a troubling paradox. While it imposes present costs on countries by requiring them to reduce emissions, it also creates valuable assets in the form of tradable emissions permits. Because limiting emissions creates a scarcity where none previously existed, such a plan allows "green revenues" to flow to the owners of the permits. The fact of wealth creation is troubling because the value of the permits might be dissipated or used for corrupt purposes. We can think of three different ways that the valuable permits could be allocated: (a) auction, (b) building political consensus, and (c) corruption. The optimal approach would be for governments to auction these permits and use the revenues for worthwhile purposes -- to lower existing taxes, reduce budget -17- deficits, or expand valuable programs. Such use would be the fairest way to dispose of the revenues and would enhance efficiency if the revenues were used to lower other taxes. If the permits are auctioned and the revenues recycled, the net impact on the country could be quite low. Given political imperfections and the need to buy political support, a government might give permits to those politically powerful groups who were adversely affected by the plan. This occurred when the U.S. instituted its sulfur emissions-trading regime, where the permits were allocated to polluters on the basis of historical emissions. This approach is defective for a number of reasons: it overcompensates polluters, it sets up terrible incentives for firms to ratchet up their emissions, and it unnecessarily increases the deadweight loss of taxation. While the plan might strongly recommend against such allocations, they would probably not constitute a violation of the agreement. The real difficulty arises with respect to non-democratic and corrupt regimes. In a polar case, we might imagine a venal dictator who sells most of the permits (perhaps to another country), pockets the proceeds, turns the screws tightly on the population by severely limiting fuel imports, and moves the assets abroad to Riviera villas and fine Bordeaux wines. This might even worsen the environmental condition of the country and world by encouraging desperate foraging for firewood and increased deforestation. As examples, consider the cases of Iraq and Nigeria, each of which had emissions of around 30 million tons of industrial CO₂ emissions in 1994. These could easily sell for $300 million or more each year of hard currency on the open market (or $3 billion annually if the environmentalist plans were imposed). We would clearly want such countries to participate, but we could not be confident that the revenues from the permits would not be used for acquiring nuclear materials or bolstering Swiss bank accounts. To avoid antisocial activities, we would probably impose conditionality requirements on the sale and use of emissions permits; for example, renegade countries (however that is defined) might not be permitted to sell permits to other countries. Of course, the more burdensome are the "ethical" restrictions on the disposition of the permits, the less attractive participation becomes for countries, so there is a delicate tradeoff here. -18- 13. Simulations Using the RICE Model: The Setup An important question is the relative efficiency of different climate-policy regimes. For this analysis, I use an updated version of the "RICE model" of global warming 10 as modified for this analysis. The RICE model is an integrated assessment model that includes output, emissions, concentrations, climate change, and climate-change damages for a number of world regions. The model uses the data for 13 regions of the world for the estimates (U.S., Europe, China, India, Russia, and 8 other groups of countries). Most of the parameters are from the original RICE model, but the data are updated to 1994 levels of output and emissions. Using the model, we can calculate both the costs of compliance and the benefits of slowing climate change in each region for the different policy proposals. The model is then run for a number of different policy proposals. A baseline is the "market" or "no-control" case where no climate policies are in place. The "optimal" policy corresponds to complete participation with lump-sum redistribution. There are three CAP regimes with three different schedules for participation. The one described above is the "middle progressivity." There are in addition a regressive and a more progressive plan.¹ 11 Finally, we investigate three proposals that correspond to plans that are currently under discussion. 10 See Nordhaus and Yang, op. cit. for a complete discussion. "The "middle progressive" CAP protocol assumes an entry-level income of $5000 per capital and full-participation income of $15,000. More precisely, the percent participation (the f function) goes linearly from 0 at a per capita income of $5000 to 100 percent at $15,000. Alternative progressivity schedules have entry levels and full participation of ($10,000 and $20,000) for a "progressive" approach and ($0 and $5000) for a "regressive" approach. For the simulations, we assume that there is perfect compliance and that countries do not join ahead of schedule. We have simplified the calculations by using the actual damage function rather than the median-allowable-emissions voting function. In addition, we assume that the economic and technological variables are known with certainty. -19- The Rio Treaty plan, which has emissions limits at 1990 levels for the OECD region. The U.S. proposal, or Rio limits with emissions trading among the OECD countries. The "Berlin mandate," which follows the Rio Treaty but has 20 percent cuts in emissions from 1990 levels starting in 2010 and does not allow trading. The basic point is to investigate the comparative performance of the different regimes. That is, what fraction of the benefits of the optimal control policy is captured by the emissions regime, and how well do the different regimes attain the ultimate goal of slowing climate change? 14. Empirical Results The enclosed figures and tables show the major results. Figures 4 and 5 shows the emissions under different plans. Figures 6 and 7 show the levels of concentration and temperature change for the different proposals. There are significant reductions in emissions, concentrations, and global temperature change in all the plans. Current approaches in the Rio and Berlin plans do relatively well for the next century or so in terms of the environmental objectives. In the longer run, they lag behind the CAP plans because they have no mechanism to bring in big developing-country emitters in the future. Under the CAP plans, all important countries (including India and China) incur reduction obligations in the next century, so most emissions are covered relatively quickly. While the environmental picture does not differ much among the different approaches, there is an enormous difference in the economics. Figure 8 shows the net benefits of the alternative plans. These measure the present value of control costs and climate damages from 1990 on, in 1994 prices, discounted at a goods discount rate of 5 percent annually (the measurement convention is that costs are negative and measured relative to no controls). Table 1 shows the basic data, with 1A showing the absolute levels and 1B showing the impacts as a share of national outputs. The bottom half of the tables show the impact relative to no controls for each region and for the world. Two fundamental findings emerge from these simulations: Economic turkey. The three current proposals on the international agenda (Rio, the U.S. modification of Rio, and the Berlin protocol) have negative net benefits, and substantial ones. These plans impose more than $100 billion of discounted costs for -20- the Rio approaches and $1 trillion of discounted costs for the Berlin approach. By contrast, the CAP proposals all have positive global benefits. The Rio-based plans suffer from limiting emissions to a small fraction of emissions, by having rigid targets, and (except for the U.S. plan) by proscribing emissions trading. Political clunker. The three Rio-style options are politically inept and cannot survive the political market place in most democracies. They impose enormous costs on the major participants - the high-income countries - and confer large benefits on those who do nothing. The U.S. fares extremely poorly under all three of the current proposals, including its own. These plans would be rejected by countries acting in their self-interests. By contrast, the CAP plans have positive or at worst small negative net benefits for all countries. Figure 9 shows the distribution of gains and losses under two plans -- the middle CAP plan and the U.S. plan of Rio targets with trading. This shows the difference in the distribution of benefits of the different approaches. Figure 10 shows the carbon emissions prices that are estimated for the different plans. The Rio-with-trading and the Berlin mandate have extremely high emissions prices, reaching $100 quickly and heading north from there very quickly. These carbon taxes are unrealistically high from an administrative and political point of view. For example, the mine-mouth price of U.S. coal would rise by between 60 and 2000 percent over current prices depending on the time period and plan. The carbon tax in the CAP plan, by contrast, starts around $6 per ton carbon and rises to $35 per ton by 2100. Table 2 summarizes the economic impacts and efficiencies of the different approaches relative to the optimal emissions reduction, which has a benchmark of 100 percent economic and environmental efficiency. The first two columns show the overall net economic impacts. The three current proposals have benefit-cost ratios well below 1. Trading improves the benefit-cost ratio of the Rio approach, but does not raise it anywhere near unity. Focusing on the long term concentration and temperature effects in the middle and right-hand columns, we see that - in addition to having much better economic efficiency - the CAP plans also have better long-term environmental benefits than the Rio- style proposals. Finally, Table 3 shows the impact of different plans on the United States. The U.S. has little net gain or loss under the CAP plans but is a major loser under the Rio- style proposals. The most important conclusions that come from these simulations are (a) that current approaches are no-starters in any negotiations in which countries take actions based on their self-interest and the (b) current approaches are highly inefficient in their design. In short, they are both bad politics and bad economics. -21- 15. Conclusions The CAP proposal described here is an attempt to flesh out a emissions-trading proposal in the context of the global public good of global warming. In principle, it satisfies most of the design criteria discussed in section 2. An efficient trading regime would have the advantage of integrating policy with an underlying cost-benefit approach that maximizes global incomes. Further, it would allow countries to arrive at a level of the public good that reflects individual country preferences. Model simulations indicate that the CAP plan would achieve between 40 and 75 percent of the potential economic benefit - and more than 96 percent of the long-term environmental benefit - depending upon the time frame and progressivity schedule. These compare with current proposals - such as the original Rio targets, the Berlin mandate, and the U.S. modification of the Rio targets to allow emissions trading - which have less favorable long-run environmental impacts and have significantly negative economic impacts. At the same time, we must register some important reservations about the proposal. Many will think that the proposal to have sanctions linked to punitive import duties for non- participants is a dangerous and unwarranted distortion of international trade. Additionally, the mechanism for determining country preferences is complicated and may seem arcane to political decision makers (although it is simple compared to many "familiar" issues such as tax or real-estate law). And the treatment of inept and corrupt countries poses particular problems. Most of these issues will, however, necessarily arise in any effective plan. Notwithstanding these difficulties, the CAP proposal or some relative has the advantage of being both economically advantageous and politically workable whereas all current proposals are politically unrealistic and economically wasteful while conveying no environmental advantage over the CAP plan. -22- Figure 4 25 Emissions 20 Emissions per year (bill t C) 15 10 Opt Mkt Rio Mid 5 1980 2000 2020 2040 2060 2080 2100 2120 2140 2160 06/17/97: 17:01:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9.wb3 Figure 5 Emissions Alternative Plans 30 Optimal Regress 25 Emissions (bill. tons C per year) Middle Progress 20 Berlin Rio w/o trade 15 Rio w/ trade Market 10 2100 2200 06/17/97: 17:10:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9.wb3 Figure 6 Concentrations Alternative Plans 3000 Optimal 2500 Regress Concentrations (bill. tons C per year) Middle Progress 2000 Berlin Rio w/o trade 1500 Rio w/ trade Market 1000 2100 2300 06/17/97: 17:01:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9.wb3 Figure 7 Temperature Alternative Plans 6.5 Optimal Regress 5.5 Middle Temperature (degrees C) Progress 4.5 Berlin Rio w/o trade 3.5 Rio w/ trade Market 2.5 2150 2350 06/17/97: 17:03:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9.wb3 Figure 8 Net Economic Impact Alternative Plans 200 0 Net impact (billions $, 1994 prices) -200 -400 -600 -800 -1000 Optimal Middle Rio w/o trade Berlin Regress Progress Rio w/ trade 06/17/97: 17:03:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9.wb3 Figure 9 Gainers and Losers Alternative Plans 4 2 Gains or losses vs. market (.01 .01 % %) 0 -2 -4 MidProg -6 Rio w/o Trade -8 USA Russia China India OBE Middle Tiny Japan FSC EU Indo/Braz BLU Small TOTAL 06/17/97: 17:04:G:\Home Disk F\RES-CLIM\RICE\Rice97\NRf#9.wb3 Figure 10 Carbon Tax Rates in Different Plans 200 Optimum $/tC Rio W Trade 100 0 2000 2050 2100 06/17/97: 17:05:G:\Home Disk F\RES-CLIM\RICE\Rice97\NR#9 wb3 I able IA Net Benefits of Alternative Trading Regimes Billions of 1994 US dollars Relative to No Global Warming Indon/ USA Japan Russia FSC China EU India Brazil OBE BLU Middle Small Tiny TOTAL Market -520.8 -336.4 -32.2 -39.3 -232.9 -695.4 -123.7 -228.7 -56.8 -104.2 -695.6 -260.0 -116.1 -3442 Optimal -504.9 -318.9 -36.5 -40.7 -231.2 -661.3 -119.9 -214.7 -57.8 -98.6 -659.7 -246.5 -109.1 -3300 Regress -520.1 -328.7 -29.3 -35.7 -218.5 -681.3 -115.7 -219.3 -51.2 -98.6 -678.6 -248.1 -110.0 -3335 Middle -525.2 -332.7 -29.8 -37.6 -222.2 -689.5 -118.5 -220.6 -53.8 -100.2- -675.1 -250.1 -111.6 -3367 Progress -523.6 -333.8 -31.0 -38.2 -225.4 -690.9 -119.9 -222.4 -55.0 -101.2 -677.8 -252.5 -112.7 -3385 Rio w/o trade -636.5 -396.1 -30.9 -37.8 -222.1 -759.1 -118.3 -219.2 -54.5 -100.0 -668.5 -249.5 -111.2 -3604 Rio w/ trade -617.5 -382.1 -31.0 -37.9 -222.4 -757.1 -118.4 -219.4 -54.6 -100.1 -669.3 -249.8 -111.3 -3571 Berlin -1057.3 -578.7 -29.7 -36.4 -213.3 -1064.5 -113.7 -210.6 -52.4 -96.1 -642.7 -239.9 -106.9 -4442 Relative to No-Control Solution Indon/ USA Japan Russia FSC China EU India Brazil OBE BLU Middle Small Tiny TOTAL Market 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 Optimal 15.8 17.4 -4.3 -1.4 1.8 34.1 3.8 14.0 -1.1 5.6 35.8 13.4 7.0 142 Regress 0.6 7.7 2.8 3.6 14.5 14.1 8.0 9.3 5.6 5.7 17.0 11.9 6.1 107 Middle -4.4 3.6 2.4 1.7 10.8 5.8 5.2 8.1 3.0 4.1 20.5 9.9 4.5 75 Progress -2.8 2.6 1.1 1.1 7.5 4.5 3.8 6.3 1.7 3.0 17.7 7.5 3.4 57 Rio w/o trade -115.8 -59.8 1.3 1.5 10.8 -63.7 5.4 9.5 2.2 4.3 27.1 10.5 4.9 -162 Rio w/ trade -96.8 -45.8 1.2 1.4 10.6 -61.7 5.3 9.3 2.2 4.2 26.3 10.2 4.7 -129 Berlin -536.5 -242.3 2.4 2.9 19.7 -369.1 10.0 18.1 4.3 8.1 52.8 20.1 9.2 -1000 Item: Discounted GDP (trillions) 191.6 111.7 12.2 6.3 38.5 236.0 16.6 32.5 8.6 15.2 107.4 38.7 16.5 831.9 06/05/97: 11:59: F:\RES-CLIM\RICE\Rice97\NR#8.wb3 Table 1B Net Benefits of Alternative Trading Regimes Percent of Discounted GDP (x 100) Relative to No Global Warming Indon/ Stand. USA Japan Russia FSC China EU India Brazil OBE BLU Middle Small Tiny TOTAL Dev. Market -27.2 -30.1 -26.4 -62.1 -60.4 -29.5 -74.5 -70.3 -65.9 -68.7 -64.7 -67.2 -70.5 -41.4 18.3 Optimal -26.3 -28.6 -29.9 -64.2 -60.0 -28.0 -72.2 -66.0 -67.1 -65.0 -61.4 -63.7 -66.2 -39.7 17.3 Regress -27.1 -29.4 -24.1 -56.4 -56.7 -28.9 -69.7 -67.5 -59.4 -64.9 -63.2 -64.1 -66.7 -40.1 17.0 Middle -27.4 -29.8 -24.4 -59.5 -57.6 -29.2 -71.3 -67.9 -62.4 -66.0 -62.8 -64.6 -67.7 -40.5 17.3 Progress -27.3 -29.9 -25.4 -60.4 -58.5 -29.3 -72.2 -68.4 -63.9 -66.7 -63.1 -65.3 -68.4 -40.7 17.5 Rio w/o trade -33.2 -35.5 -25.3 -59.7 -57.6 -32.2 -71.2 -67.4 -63.3 -65.9 -62.2 -64.5 -67.5 -43.3 15.6 Rio w/ trade -32.2 -34.2 -25.4 -59.8 -57.7 -32.1 -71.3 -67.5 -63.3 -65.9 -62.3 -64.6 -67.6 -42.9 15.9 Berlin -55.2 -51.8 -24.4 -57.5 -55.3 -45.1 -68.5 -64.8 -60.8 -63.3 -59.8 -62.0 -64.9 -53.4 11.0 Relative to No-Control Solution Indon/ Stand. USA Japan Russia FSC China EU India Brazil OBE BLU Middle Small Tiny TOTAL Dev. Market 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Optimal 0.8 1.6 -3.5 -2.1 0.5 1.4 2.3 4.3 -1.3 3.7 3.3 3.5 4.2 1.7 2.4 Regress 0.0 0.7 2.3 5.7 3.8 0.6 4.8 2.9 6.5 3.8 1.6 3.1 3.7 1.3 1.9 Middle -0.2 0.3 1.9 2.6 2.8 0.2 3.1 2.5 3.5 2.7 1.9 2.6 2.8 0.9 1.1 Progress -0.1 0.2 0.9 1.7 1.9 0.2 2.3 1.9 2.0 2.0 1.6 1.9 2.1 0.7 0.8 Rio w/o trade -6.0 -5.4 1.0 2.3 2.8 -2.7 3.2 2.9 2.6 2.8 2.5 2.7 3.0 -1.9 3.2 Rio w/ trade -5.0 -4.1 1.0 2.3 2.7 -2.6 3.2 2.8 2.5 2.7 2.4 2.6 2.9 -1.5 2.8 Berlin -28.0 -21.7 2.0 4.6 5.1 -15.6 6.0 5.6 5.0 5.4 4.9 5.2 5.6 -12.0 11.6 06/05/97: 11:59: F:\RES-CLIM\RICE\Rice97\NR#8.wb3 Table 2 Efficiency Economic Temperature* Concentrations* Percent Benefit/Cost 2150 2350 2100 2300 Regime of Optimum* Ratio** Market 0 0.00 0.0 0.0 0.0 0.0 Optimal 100 2.88 100.0 100.0 100.0 100.0 Regress 75 2.87 96.2 99.6 93.0 99.8 Middle 53 2.87 82.7 98.5 70.7 98.8 Progress 40 2.87 61.9 96.3 46.3 96.5 Rio w/o trade -114 0.44 67.7 41.0 73.8 44.2 Rio w/ trade -91 0.49 69.4 40.1 72.8 43.6 Berlin -704 0.20 97.3 51.9 111.5 55.5 * Percentage gain in net benefits, temperature, or concentrations as percent of optimal relative to market. **Ratio of reduction in climate damages to abatement costs. 06/05/97: 14:12: F:\RES-CLIM\RICE\Rice97\NRff9.wb3 Table 3 Costs and Benefits for the United States [Billions of dollars, 1994 prices; costs and damages measures negatively*] Relative to no controls Costs Benefits Net Benefits Benefit-Cost Ratio Market 0.0 0.0 0.0 na Optimal -14.0 29.9 15.8 2.13 Regressive -20.8 21.4 0.6 1.03 Middle -19.4 14.9 -4.4 0.77 Progressive -14.3 11.5 -2.8 0.80 Rio w/o Trading -131.9 16.2 -115.8 0.12 Rio with Trading -112.3 15.6 -96.8 0.14 Berlin with 20 percent cuts -571.1 34.6 -536.5 0.06 * Calculations are from the 1997 version of the RICE model. All numbers cover the period 1990 through 2400 and include the present value of the costs of compliance (including both resource costs and permit purchases) and the climate damages discounted at a 5 percent discount rate back to 1990 in 1994 prices. The "Costs" column measures the compliance costs and the "Benefits" column measures the averted climate damage. 06/16/97: 07:52: F:\RES-CLIM\RICE\Rice97\NR#8a.wb3 DRAFT On Stabilizing CO2 Concentrations - Cost-Effective Emission Reduction Strategies Alan Manne, Stanford University Richard Richels, Electric Power Research Institute February 25, 1997 This research results from our involvement in Stanford University's Energy Modeling Forum 14 Study. For presentation at the IPCC Asia-Pacific Workshop on Integrated Assessment Models, the United Nations University, Tokyo, Japan, 10-12 March 1997. We are indebted to David Chang and Robert Parkin for research assistance. We have benefited from discussions with Sharmila Barathan, Jae Edmonds, Howard Gruenspecht, Eric Haites, Henry Jacoby, William Nordhaus, Stephen Peck, Leo Schrattenholzer, John Weyant, Tom Wigley, and ZhongXiang Zhang. Funding was provided by the Electric Power Research Institute. The views presented here are solely those of the individual authors. Abstract With the adoption of the Berlin Mandate, developed countries are being asked to set emission limits for the early decades of the next century. The size of the reductions is currently the subject of international negotiations. This paper is intended to contribute to the analysis and assessment phase leading up to the adoption of new targets and timetables. However, we take a somewhat different approach than that suggested by the Berlin Mandate. Rather than focus exclusively on the next steps by developed countries, we view the issue from the perspective of the Convention's ultimate objective, the stabilization of atmospheric concentrations. We examine what might constitute cost-effective strategies for limiting CO2 concentrations to alternative levels. We then explore the implications for near-term mitigation decisions and for long-term participation by the developing countries. 1. Introduction In recent years, global climate change has become one of the most contentious environmental issues facing the international community. The UN Framework Convention calls for the "stabilization of greenhouse gases in the atmosphere at a level that would prevent dangerous anthropogenic interference with the climate system."¹ Yet the issue of what constitutes "dangerous anthropogenic interference" is likely to remain the subject of intense scientific and political debate for some time. For the present, international negotiations must remain an ongoing process -- with ample opportunities for learning and for midcourse corrections. We are currently in the midst of one such review cycle. When initially put forward at the "Earth Summit" in 1992, the Framework Convention called upon developed countries to aim to return emissions to 1990 levels by the year 2000. At the first meeting of the Conference of the Parties in 1995, these commitments were deemed inadequate. As a result, the so-called "Berlin Mandate" was adopted. This called upon developed countries to set "quantified limitation and reduction objectives" for the post-2000 time frame.² Although the Berlin Mandate is explicit in its call for additional reductions, it does not specify how large the reductions should be. Rather it specifies an "analysis and assessment" phase to help inform the decision making process. The deadline for new commitments is December 1997. A wide variety of proposals have been put forward in anticipation of this deadline. These proposals range from sharp cuts in near-term emissions to a more gradual transition away from carbon-intensive fuels. The international research community is actively engaged in trying to understand the environmental and economic implications of these policy proposals. This paper is intended to contribute to the process of analysis and assessment. However, we take a somewhat different approach than that suggested by the Berlin Mandate. Rather than focus exclusively on the next steps by developed countries, we view the issue from the perspective of the Convention's ultimate objective, the stabilization of atmospheric concentrations. We examine what might constitute cost-effective strategies for limiting CO2 concentrations to alternative levels. We then explore the implications for near-term mitigation decisions and for long-term participation by the developing countries. There are several reasons why a broader perspective is desirable. The Intergovernmental Panel on Climate Change (IPCC) has demonstrated that if CO2 concentrations were to be stabilized at any of the levels it examined, this would require an eventual and sustained reduction in emissions to substantially below current levels. 3 Developed countries cannot do the job by themselves. Nor can the transition be accomplished overnight. Cost-effective strategies will require both a global and a long-term perspective. Any analysis of stabilization must confront the divisive issue of burden sharing. With trade in carbon emission rights, emission reductions can be made where it is cheapest to do so, regardless of their geographical location. The allocation of permits will have little impact on the least-cost global strategy. 4 It will, however, have profound affects on who pays. Consistent with the Framework Convention, we adopt a burden sharing scheme that initially places the onus on developed countries. We then discuss the implications for international negotiations. Economic analysis can play an important role in the climate debate. It can help policy makers identify least-cost mitigation strategies from a global perspective. In doing so, this helps to minimize the size of the overall burden. It can shed light on the implications of alternative burden sharing schemes at the regional level. Economic analysis, however, cannot tell us how a given burden should be allocated. Fairness and equity issues must necessarily be left to the international negotiation process. Finally, we emphasize that our analysis is confined to mitigation costs. We recognize that this is not the whole story, but it is an important part. Article 3 of the Framework Convention states that "policies and measures to deal with climate change should be cost-effective so as to insure global benefits at the lowest possible costs. Identifying least-cost mitigation strategies can free up valuable resources for addressing alternative uses. 2. The model The analysis is based on MERGE - a model for evaluating the regional and global effects of greenhouse gas reduction policies. MERGE provides a bottom- up representation of the energy supply system. For a given scenario, a least- cost choice is made among specific technologies for the generation of electricity and for the production of nonelectric energy. As fossil fuels (coal, oil and gas) are exhausted, their prices rise and carbon-free alternatives become more competitive. To allow for inertia in the energy supply system, decline and expansion constraints are placed on existing and new technologies, respectively. A top-down perspective is taken for the balance of the economy. These sectors are modeled through nested constant elasticity of substitution production functions. The production functions determine how aggregate output depends upon the inputs of capital, labor, electric and nonelectric energy. In 2 this way, the model allows for both price-induced and autonomous (non- price) energy conservation and also for interfuel substitution. A "putty-clay" formulation is used to allow for the lags in adapting to changes in energy prices. In MERGE, the savings and investment process is affected by intertemporal and interregional forces. Each region is represented as though it maximizes discounted utility (the logarithm of consumption) subject to an intertemporal budget constraint. Its wealth includes not only capital, labor, and exhaustible resources, but also its negotiated share in global carbon emission rights. With this objective function, the costs of abatement are defined as the losses in the discounted value of consumption associated with alternative carbon constraints. In previous versions of MERGE, the world was subdivided into five geopolitical regions.⁶,⁷ The present version of the model, known as MERGE 3.0, divides the world into nine regions: 1) the USA, 2) OECDE (Western Europe), 3) Japan, 4) CANZ (Canada, Australia and New Zealand), 5) EEFSU (Eastern Europe and the former Soviet Union), 6) China, 7) India, 8) MOPEC (Mexico and OPEC) and, 9) ROW (the rest of world). The further disaggregation provides better alignment with the Annex 1/non-Annex 1 structure of the Framework Convention. It provides more details concerning winners and losers under alternative burden sharing schemes, and it distinguishes between the major oil importing and exporting regions. Population trends for each region are taken as exogenous. Per capita incomes are determined primarily by the rate of labor force productivity. Between 1990 and 2020, our projections are consistent with the conventional wisdom median values of the International Energy Workshop poll. 8 For the world as a whole, GDP growth is projected at an average annual rate of 2.5% between 1990 and 2100. It is assumed that there are ultimate limits to economic growth, and that there will eventually be convergence between the per capita incomes in the OECD countries and those in the rest of the world. Figure 2.1 shows our specific projections of per capita GDP in each of the nine regions. MERGE is based on a general equilibrium formulation of the global energy- economic system. This enables us to model trade in oil, gas and carbon emission rights. The model does not, however, account for the effect of an economic slowdown in one region on the full range of exports of another. It may therefore be ignoring some important "spillover" effects. MERGE is not designed to address short-run macroeconomic issues such as unemployment and inflation. The employment level is exogenous, and there are instantaneous adjustments to policy shocks. As a result, the model may overlook some costly short-term dislocations. 3 Figure 2.1 Per Capita GDP 180 160 140 USA 120 OECDE Thousands of US Dollars Japan 100 CANZ EEFSU CHINA 80 INDIA MOPEC 60 ROW 40 20 0 1990 2050 2100 2150 2200 Figure 2.2 Regional Carbon Emissions 20 18 16 14 12 Billion tons of carbon non-Annex 1 10 EEFSU OECD 8 6 4 2 0 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Figure 2.3 Total Primary Energy Use - Basecase 100% 80% Carbon-free 60% Gas Oil 40% Coal 20% 0% 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 CO2 mitigation costs are determined by 1) the emissions baseline -- i.e., how emissions are apt to grow in the absence of policy interventions, 2) the cost and availability of alternative supply and demand-side options, and 3) the magnitude of the CO2 constraint. For the present analysis, several supply and demand parameters of the energy-economy submodel have been adjusted so that the baseline tracks the IPCC IS92a⁹ scenario through the year 2100. Figure 2.2 shows carbon emissions for the OECD, EEFSU and non-Annex 1. Figure 2.3 shows the corresponding total primary energy use by fuel type. Some observers have suggested that the exogenous specification of technical change will overstate the costs of a carbon constraint. They argue that an international carbon abatement agreement will automatically induce innovations in carbon-saving technologies. We do not share their optimism on the automatic nature of such innovations. (Consider, for example, the history of both fission and fusion technologies.) We do believe, however, that carbon constraints might speed up the process of technology diffusion. MERGE 3.0 incorporates the notion of "endogenous technology diffusion". Specifically, in the electric power sector, the near-term adoption of high cost carbon-free substitutes makes it possible to introduce low-cost alternatives more rapidly in the future. Upon request, the authors will supply computer files that fully document the assumptions underlying the model.¹ 3. Scenario design We focus on three factors critical to determining the costs of stabilizing concentrations at a particular level: the choice of global emissions pathway, the degree of international cooperation and the burden sharing scheme. In this section, we describe our assumptions with regard to each. The global emissions path to stabilization. In 1994, Working Group I (WG1) of the Intergovernmental Panel on Climate Change (IPCC) published a set of concentration profiles for stabilizing atmospheric CO2 at 350, 450, 550, 650 and 750ppmv. 10 The purpose of their estimates was to illustrate what might be required in terms of global CO2 emissions reductions in order to stabilize concentrations at these different levels. Subsequently, Wigley, Richels and Edmonds (WRE) published an alternative set of emission profiles for achieving the WG1 concentration targets. 11 Although WG1 and WRE are identical in terms of the prescribed stabilization levels and attainment dates, they differ in the routes to stabilization (Figure 3.1a). The IPCC, while not taking a position on the desirability of one set over another, published both in their 1995 scientific assessment. 12 1 Contact [email protected] 7 Figure 3.1 Alternative Routes to CO₂ Stabilization 800 750ppmv 700 650 600 550 500 450 ppmv 400 300 WRE 200 WG1 100 o 1990 2050 2100 2150 2200 a) Alternative concentration profiles 20 IS92a 18 WRE WG1 16 14 12 Billion tons of carbon 10 8 6 750ppmv 650 4 550 2 450 0 1990 2050 2100 2150 2200 b) Alternative fossil fuel emisson paths Figure 3.1b shows the emission rates required to achieve stabilization via the WG1 concentration profiles (the dashed lines) and the WRE concentration profiles (the solid lines). The calculations were made using the Wigley carbon cycle model. The WRE curves were constructed so that they would follow the central IPCC "existing policies" or "baseline" emissions scenario (IS92a) in the early years. The higher the stabilization target, the longer the adherence to IS92a. In contrast, the WG1 curves depart from IS92a immediately. WRE assert that concentration pathways with higher near-term emissions are apt to have lower mitigation costs. They cite several economic studies that have examined how mitigation costs might vary with the timing of emission reductions. 13,14,15,16 These studies suggest that the time path to stabilization may be as important as the concentration target itself in determining the overall discounted costs. They conclude that emission pathways that provide for a gradual transition away from fossil fuels are apt to be less expensive in terms of mitigation costs. The WRE analysis is primarily qualitative. While drawing upon other studies to make their points, no explicit analysis is made of the mitigation costs associated with either the WG1 or WRE pathways. This is the issue to which we now turn. In this paper, we examine the costs of stabilizing concentrations at 450, 550, 650, and 750ppmv, first following the WG1 emissions pathway and then those suggested by WRE. In each case, we examine the costs to Annex 1 and non-Annex 1 countries under alternative burden sharing schemes. The extent of international cooperation. A number of studies have shown that the marginal costs of emissions abatement might vary considerably among regions.¹⁷ This will be particularly the case in those periods when emission reductions are confined to Annex 1 countries. Clearly, it is inefficient to incur high marginal domestic abatement costs in Annex 1 countries when the marginal cost of emissions abatement is lower in non- Annex 1 countries. It is equally clear that it would be unrealistic to expect the non-Annex 1 countries to bear the burden of domestic reductions so as to achieve a globally cost-effective result. This suggests opportunities for efficiency gains through various forms of "joint implementation". This could be done on a bilateral project-by-project basis during the earlier years of an international agreement. Over the long term, however, it is more promising to explore market mechanisms such as a system of international allocations of tradable carbon emission rights. Here we first examine mitigation costs when emission reductions are confined to the region of origin. We then calculate the benefits from international cooperation using trade in emission rights as a proxy for other forms of cooperative mechanisms with side payments. 9 When we assume that reductions take place wherever it is cheapest to do so (regardless of the geographical location), we refer to this as "interregional" or "where" flexibility. When there is a choice in the timing of emission reductions, we refer to this as "intertemporal" or "when" flexibility. The burden sharing rule. Region-specific mitigation costs will also depend upon how emission reductions are allocated among regions. Consistent with the Berlin Mandate, we assume that the burden will fall on Annex 1 countries during the initial decades of an agreement. During this period, Annex 1 countries would be required to limit their emissions to amounts proportional to their 1990 levels. Even if the Annex 1 countries were to reduce their emissions to zero, this would not be sufficient to limit global concentrations. Eventually, the non- Annex 1 countries will also have to limit their emissions. The ultimate concentration target will affect the date at which these non-Annex 1 countries must begin to participate in a global agreement. The more ambitious the target, the sooner they will have to participate in such an agreement. Once the non-Annex 1 countries do agree to a constraint, however, it is plausible to assume that there will be a gradual transition to equal per capita emission rights. (For one such scheme, see Table 3.1.) Table 3.1 Structure of Burden Sharing Scheme Concentration target Date at which non- Date by which transition (ppmv) Annex 1 countries must to equal per capita begin to limit emissions emission rights is achieved 450 2020 2040 550 2030 2050 650 2040 2060 750 2050 2070 With "where" flexibility, global mitigation costs are independent of the burden sharing scheme. That is, reductions will take place wherever it is cheapest to do so regardless of the geographical location. This is not the case, however, when emission reductions are restricted to the region of origin. With this type of restriction, the burden sharing scheme will affect both the global and the regional costs. 10 4. Global costs of stabilization at 550ppmv Mitigation costs are incurred when the imposition of a carbon constraint leads to a reallocation of resources from the pattern that would be preferred in the absence of the constraint. A carbon constraint will lead to more expensive conservation activities and to fuel switching. There are changes in both domestic and international prices. In most cases, these forced adjustments lead to a reduction in economic performance. The tighter the constraint, the greater the effect. With MERGE, we can calculate how mitigation costs vary with the choice of concentration profile. At the present time, there is little consensus on what constitutes an appropriate concentration target. There is even less consensus on the choice of a pathway to stabilization. We shall begin by focusing on a concentrations target of 550ppmv -- approximately twice the preindustrial level. Later, we will explore the implications of adopting alternative concentration targets. Figure 3.1b shows two sets of emission pathways for stabilizing concentrations at 550ppmv. The burden sharing rule (Table 3.1) will determine how emissions might be apportioned between regions. The results are summarized for three broad groups of countries: 1) the OECD (USA, Western Europe, Japan, Canada, Australia and New Zealand), 2) EEFSU (Eastern Europe and the former Soviet Union) and 3) developing countries. In the language of the Framework Convention, the first two groups are described as Annex 1 countries. All others (the developing countries) are non-Annex 1 countries. Figure 4.1 shows the implications of the burden sharing rule for these three regional groupings. The figure provides some insight into both global and regional costs. Under the WRE scenario, Annex 1 countries have some room for emissions growth, at least during the early decades of the 21st century. This is not the case, however, under the WG1 scenario. Here, Annex 1 emission reductions must begin immediately. This decline would be inconsistent with post-1990 trends in all but a few countries. Figure 4.2 compares mitigation costs over the 21st century. Consumption losses are expressed in constant dollars, discounted to 1990 at 5% per year. Notice that costs are considerably lower for the WRE pathway. There are several reasons why this turns out to be the case. A concentration target defines an emissions budget, i.e., an allowable amount of carbon to be released between now and the date at which the target is to be achieved. A cost-effectiveness analysis is focused upon how this global budget might be allocated over time. 11 Figure 4.1 Alternative Emission Pathways for Stabilizing Concentrations at 550ppmv 10 9 8 Total 7 6 non-Annex 1 Billion tons of carbon 5 4 3 X 2 OECD 1 EEFSU 0 1990 2050 2100 2150 2200 a) WG1 10 9 Total 8 7 6 Billion tons of carbon 5 non-Annex 1 4 3 X 2 OECD 1 EEFSU 0 1990 2050 2100 2150 2200 b) WRE Figure 4.2 Regional Costs of Stabilizing Concentrations at 550ppmv -- discounted to 1990 at 5% 10 8 non Annex-1 6 EEFSU $ OECD 4 Trillions of dollars 2 0 w/o w/o "where" "where" "where" "where" flexibility flexibility flexibility flexibility -2 WG1 WRE -4 Shifting emission reductions into the future provides valuable time for: 1) adapting the energy using and energy producing capital stock, 2) developing low cost substitutes to carbon intensive fuels, and 3) removing carbon from the atmosphere via the carbon cycle. In addition, with the economy yielding a positive return on capital, future reductions can be made with a smaller commitment of today's resources. For a more detailed discussion of these factors, see ref. (11). Figure 4.2 also estimates the benefits from international cooperation. Without "where" flexibility, the immediate emission reductions are confined directly to the Annex 1. The more countries that participate in an international agreement, the greater become the opportunities for cost- effective trades. It then becomes possible for countries with high marginal abatement costs to purchase emission rights from countries with low marginal abatement costs. From a global perspective, combining "where" flexibility with a more gradual transition away from fossil fuels substantially reduces the present value of mitigation costs. It turns out that there can be cost reductions as high as 90% when we combine both types of flexibility. (Compare the leftmost to the rightmost bar in Figure 4.2.) The discounted cost savings to the international community appear to be of the order of trillions of dollars over the 21st century. This is consistent with earlier studies which focused exclusively on 18,19 near-term targets and timetables. Table 3.1 describes a burden sharing rule that is particularly favorable to the non-Annex 1 countries during the early decades of the next century. This is why the mitigation costs are lowest for non-Annex 1 countries under the WG1 pathway when we allow for "where" flexibility. Indeed, the WG1 emission constraints creates a sufficiently high price for emission rights and high wealth transfers during the early decades of the coming century so that the non-Annex 1 countries are actually better off in the presence of a carbon constraint than in its absence. This particular result should not, however, obscure the fact that from a global perspective, costs are far lower under the WRE pathway. 5. Annual losses when stabilizing concentrations at 550ppmv Figure 4.2 summarizes abatement costs in terms of discounted present value -- summing over all time periods. Additional insights can be gained by looking at how losses might evolve over time. In MERGE, we adopt consumption as our welfare measure. Relative impacts are more apparent when we measure annual losses in percentage rather than absolute dollar terms. 14 Figure 5.1 International Price of Crude Oil 70 60 50 40 $/barrel 30 reference case 20 WRE ("where" flexibility) WRE (w/o "where" flexibility) WG1 ("where" flexibility) 10 WG1 (w/o "where" flexibility) 0 2000 2010 2020 2030 2040 2050 c) non-Annex 1 2. 1. 2100 0602 2080 2070 0902 2050 2040 2030 OTT 2019 0 - 2 % loss C , S 9 & EEFSU (q 2- 1. 2100 0602 0802 2070 2060 2050 2040 2030 2020 2010 2000 0 * 1 = % loss C , S 9 L OCCD (e 2- 1. 2100 2090 0809 2070 0902 2050 2040 2030 2020 2010 2000 0 * - 2 % loss E , WRE ("where" flexibility) WRE (w/o "where" flexibility) S WG1 ("where" flexibility) WG1 (w/o "where" flexibility) 9 L Figure 5.2 Annual % Consumption Losses In order to explain the pattern of annual losses, it is first necessary to say something about the impact of a carbon constraint on world oil prices. A carbon constraint would have roughly the same consequence as monopsonistic cartel behavior on the part of oil importing nations. Figure 5.1 shows the international price of oil for the reference case and under the alternative pathways for stabilizing concentrations at 550ppmv. Note that oil prices are quite sensitive to the pathway to stabilization. With a tight near- term constraint, there is a drop in the international demand for crude oil. This has a dramatic effect on oil prices. There can be a differential of as much as $20 per barrel between the reference and the WG1 cases. This has important implications for the costs of a carbon constraint to both oil exporting and oil importing countries. Figure 5.2 compares annual welfare losses across scenarios for each of the three broad regional groupings. Under WG1, the Annex-1 countries must begin reducing emission immediately. Losses are highest when there is no opportunity for trading emission rights. Losses rise to more than 3% and 6% of annual consumption for the OECD and EEFSU, respectively. As a major importer of oil, the OECD benefits from the decline in world oil prices. Hence, its losses are partially mitigated. EEFSU, on the other hand, is a substantial exporter. As a result, it will be adversely affected by a drop in world oil prices, and this compounds its losses from a near-term carbon constraint. The non-Annex 1 region consists of both oil exporting and oil importing countries. Under the burden sharing scheme described in Table 3.1, the WG1 proposal leads to substantial net benefits in the early years, particularly with "where" flexibility. A tight near-term global emissions constraint would create a large demand for emission rights in Annex 1 countries. Since non- Annex 1 countries have carbon allocations up to their baseline emissions, it is in their interest to engage in domestic abatement, and to sell some of their rights to Annex 1 countries. With the parameters employed in this version of MERGE, there is a net benefit to the non-Annex 1 countries, even without "where" flexibility. The oil importers gain more than the oil exporters lose from the decline in world oil prices. Eventually, however, the region will become a net loser unless it is able to sell emission rights to Annex 1 countries. 6. The least-cost mitigation pathway -- 550ppmv The WG1 emission pathways were meant to be purely illustrative. No attempt was made to determine whether they represented an efficient transition away from fossil fuels. WRE, on the other hand, drew upon the insights of earlier studies in constructing their emission pathways. They 17 Figure 6.1 Alternative Emission Pathways for Stabilizing Concentrations at 550ppmv 20 18 16 14 baseline least-cost 12 Billion tons of carbon WG1 WRE 10 8 6 4 2 0 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Figure 6.2 Global Costs of Stabilizing Concentrations at 550ppmv -- discounted to 1990 at 5% 3.5 3 2.5 2 Trillions of dollars 1.5 1 0.5 0 WG1 WRE least-cost Figure 6.3 Value of Carbon Emission Rights with Alternative Pathways for Stabilizing Concentrations at 550ppmv 300 250 200 $/ton of carbon WG1 150 WRE least-cost 100 50 0 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 argued that allowing more time for the transition would lower mitigation costs. They did not attempt to quantify the savings from choosing one path over another. Nor did they try to identify the least-cost mitigation pathway. In the preceding sections, we analyzed the mitigation costs associated with the WG1 and WRE pathways. We now turn to the question of what might constitute a least-cost mitigation pathway for stabilizing concentrations at 550ppmv. For these calculations, we use MERGE 3.0. Rather than apply a carbon constraint derived through inverse calculations with the Wigley carbon cycle model, we now place a constraint on atmospheric concentrations and use the model to identify the least-cost mitigation pathway. Note from Figure 6.1 that the least-cost and WRE pathways for stabilizing concentrations at 550ppmv lie fairly close together, at least in the early years. That is, they tend to follow the emissions baseline during the first decade of the next century and then depart gradually. Figure 6.2 shows the results in terms of discounted present value. In each case, we assume trade among all regions. As would be expected from the previous figure, the WRE and least- cost cases are also close in terms of costs. The cases differ dramatically, however, in terms of the value of carbon emission rights. From Figure 6.3, we see that the least-cost path starts off at a low price (approximately $2/ton of carbon) and rises gradually over time. The WRE case leads to an erratic time path of carbon prices. Indeed, in some years there is an excess of emission rights. As a result, their value falls to zero. In other years, their value exceeds that of the least-cost optimal emissions reduction path. 7. The costs of stabilizing concentrations at alternative levels The selection of the 550ppmv target was purely arbitrary and not meant to imply an optimal concentration level. Given the present lack of consensus on what constitutes "dangerous" interference with the climate system, it is important to understand how mitigation costs might vary with alternative concentration targets. Figure 7.1 summarizes the results of the MERGE analysis. As would be expected, mitigation costs are a declining function of the stabilization target. Recall that a concentration target places an upper limit on the amount of CO2 to be released into the atmosphere between now and the date at which the target is to be achieved. This in effect defines an emissions budget. The lower the target, the smaller the emissions budget. Figure 7.2a shows the constraint on Annex 1 emissions during the initial decades of the next century under the WG1 scenario. In the absence of 21 Figure 7.1 Mitigation Costs for Stabilizing Concentrations at Alternative Levels -- discounted to 1990 at 5% 16 14 WG1 (w/o "where" flexibility) WG1 ("where" flexibility) 12 WRE (w/o "where" flexibility) 10 Trillions of dollars WRE ("where" flexibility) 8 6 4 2 0 450 550 650 750 ppmv WRE (q 2030 2020 2010 20000 1990 0 1 450ppmv Z E Billion tons of carbon to Awddoss S Awddoss '8 1522 9 Ma (e 2030 2020 2010 20000 1990 0 1 450ppmv 550ppmv N Awddos9 E Billion tons of carbon X to * * * S 1522 * 9 Alternative Concentration Targets Figure 7.2 Annex 1 Emission Constraints under "where" flexibility, this becomes an effective upper bound on emissions. The figure provides useful insights into the shape of the abatement cost curve. Relative to higher targets, 450ppmv implies that: 1) more carbon must be removed from the energy system, 2) there is greater need to reconfigure the existing energy using and energy producing capital stock, 3) low-cost substitutes are likely to be available in a less timely manner, and 4) there is less opportunity for discounting to reduce the present value of mitigation costs. As the concentration constraint is relaxed, each of these factors acts to lower costs. The costs of complying with WG1 are substantially reduced when we allow for "where" flexibility. Annex 1 countries are able to purchase lower marginal cost abatement alternatives from non-Annex 1 countries. As a result, the need is not as intense for early reductions. WRE produces the lowest mitigation cost possibilities. The asymmetry in the cost function between 450 and 550ppmv suggests that even with WRE, the low target will provide insufficient time to adapt the existing capital stock. From Figure 7.2b, note that a 450ppmv target would require a departure from the baseline during the first decade of the next century, and there would be an even more rapid departure thereafter. With targets of 550ppmv and above, there is time for a more gradual transition away from carbon-intensive fuels. 8. The choice of near-term mitigation strategy Figure 7.2b provides some useful guidance for the design of near-term emission strategies. If it is certain that the target is 550ppmv or above, the near-term emissions path appears to be quite robust. That is, it adheres fairly closely to the emissions baseline through 2010. It should be noted, however, that even in this case, there is some transition away from the world's current heavy dependence on carbon-intensive technologies prior to 2010. That is, inexpensive alternatives (e.g., renewables and cost-effective conservation) are introduced in increasing amounts - both on the supply and demand sides of the energy sector. However, if these alternatives are economically attractive in their own right, they will be adopted in the absence of climate policy. Suppose, on the other hand, that one believes there is some probability that the target is in the 450 to 550ppmv range. A more aggressive departure from the emissions baseline will be required. The degree of hedging depends upon the probabilities and the relative costs of two types of errors in the design of future capital stocks. That is, one must balance the risks of investing in capital stocks that lead to carbon emissions that are either too high or too low. Of course, the choice of emission pathway for meeting a prescribed concentration target must also involve consideration of the environmental 24 consequences of adopting one emission trajectory over another. The WRE emission pathways result in higher concentrations in the years preceding the date by which the target is to be achieved. For the 550ppmv case, the higher concentrations lead to pathway related differentials of up to .2 degrees C in global mean temperature and 4cm in global mean sea level change (ref. 11). To the extent that this leads to higher environmental damages, these need to be balanced against the benefits from reduced mitigation cost. 9. Some concluding comments The above analysis suggests that a more gradual transition away from fossil fuels is likely to be less expensive in terms of mitigation costs. This should not be interpreted as suggesting a "do nothing" or "wait and see" strategy. Mitigation may mean action, but action does not necessarily mean mitigation. As pointed out in the IPCC 1995 Report²⁰, climate policy requires a portfolio of responses. The challenge facing today's policy makers is to arrive at a prudent hedging strategy in the face of climate-related uncertainties. Among the options are immediate reductions of greenhouse gas emissions, investments in actions to assist human and natural systems adapt to climate change should it occur, continued research to reduce uncertainties about how much change will occur and what effects it will have, and R&D on energy supply and end-use technologies to reduce the costs of limiting greenhouse gas emissions. The issue is not one of either-or but one of finding the right blend of options. Policy makers must decide how to divide greenhouse insurance dollars among these competing needs. The present analysis has provided some useful insights bearing on this decision. Deep near-term reductions are apt to be costly. They provide less time to adapt the existing capital stock. There will be more opportunities for reducing emissions cheaply as the current capital equipment turns over. Indeed, the 1995 IPCC report states that "implementing emission reductions at rates that can be absorbed in the course of normal capital stock turnover is likely to be cheaper than forcing premature retirement now. Fortunately, with regard to carbon dioxide, the issue is one of cumulative rather than year-by-year emissions. This means that we can allow for an economical turnover of the existing capital stock if we are prepared to make sharper reductions in the future. 25 This brings us to the issue of R&D. Sharper reductions in the future will be less problematic if we can lower the costs of fuel switching and conservation. Indeed, studies by Stanford University's Energy Modeling Forum suggest that the development of economically competitive alternatives to conventional fossil fuels could substantially reduce the costs of a carbon constraint. 22 Although virtually all parties in the debate recognize the value of R&D, we have yet to develop a technology strategy for dealing with global climate change. How much should we be investing today to ensure ample supplies of low-cost alternatives in the future? What should be the nature of these investments, who would make them, and how would they be managed? Given the size of the stakes, surprisingly little attention has been devoted to these questions. 26 References 1 Intergovernmental Negotiating Committee for A Framework Convention on Climate Change (1992). Fifth Session, Second Part, New York, 30 April-9 May. 2 United Nations Climate Change Bulletin, Issue 7, 2nd Quarter 1995, published by the interim secretariat for the UN Climate Change, Convention, Geneva. 3 IPCC (1994). Climate Change 1994, Cambridge University Press. 4 Coase, R. (1960). "The Problem of Social Cost." Journal of Law and Economics, 3. 5 Intergovernmental Negotiating Committee for A Framework Convention on Climate Change (1992), op. cit. 6 Manne, A., Mendelsohn, R. and Richels, R. (1995). "MERGE: A Model for Evaluating Regional and Global Effects of GHG Reduction Policies." Energy Policy, 23 (1). 7 Manne, A. and Richels, R. (1995). "The Greenhouse Debate: Economic Efficiency, Burden Sharing and Hedging Strategies." The Energy Journal, 16 (4). 8 Manne, A. and Schrattenholzer, L. (1996). "International Energy Workshop." International Institute for Applied Systems Analysis, Laxenburg, Austria, January. 9 IPCC (1992). Climate Change 1992, Cambridge University Press. 10 IPCC (1994). Climate Change 1994, op. cit. 11 Wigley, T., Richels, R. and Edmonds, J. (1996). "Economic and Environmental Choices in the Stabilization of Atmospheric CO2 Concentrations," Nature, Vol. 379, 18 January. 12 IPCC (1996). Climate Change 1995, Cambridge University Press. 13 Nordhaus, W. (1979). The Efficient Use of Energy Resources. Yale University Press, New Haven. 14 Manne, A. and Richels, R. (1995), op. cit. 15 Richels, R. and Edmonds, J. (1995). "The Economics of Stabilizing Atmospheric CO2 Concentrations." Energy Policy, 23 (4/5). 16 Kosobud, R., Daly, T., South, D. and Quinn, K. (1994). "Tradable Cumulative CO2 Permits and Global Warming Control," The Energy Journal, 15 (2). 17 IPCC (1996). Climate Change 1995. 18 Manne, A. and Richels, R. (1996). "The Berlin Mandate: The Cost of Meeting Post-2000 Targets and Timetables." Energy Policy, 24 (3). 19 Richels, R., Edmonds, J., Gruenspecht, H. and Wigley, T. (1996). "The Berlin Mandate: The Design of Cost Effective Mitigation Strategies." EMF-14, Working Paper 3, Stanford University, Stanford, CA. 20 IPCC (1996). Climate Change 1995, op. cit. 21 IPCC (1996). Climate Change 1995, op. cit. 22 Energy Modeling Forum (1993). "Reducing Global Carbon Emissions: Costs and Policy Options." EMF-12, Stanford University, Stanford, CA. 27 A Regional Dynamic General-Equilibrium Model of Alternative Climate-Change Strategies pp. :change ge Rate By WILLIAM D. NORDHAUS AND ZILI YANG* versity, Most analyses treat global warming as a single-agent problem. The present study change presents the Regional Integrated model of Climate and the Economy (RICE) of Eco- model. By disaggregating into countries, the model analyzes different national 9-82. strategies in climate-change policy: pure market solutions, efficient cooperative lerlind, outcomes, and noncooperative equilibria. This study finds that cooperative pol- wedish icies show much higher levels of emissions reductions than do noncooperative urnal, strategies; that there are substantial differences in the levels of controls in both )-79. the cooperative and the noncooperative policies among different countries; and Esti- that high-income countries may be the major losers from cooperation. (JEL H41, e Eu- Q4, Q2, Q20) se of tional atility, Although the issue of greenhouse warming checked, recent surveys indicate that over the bility: was first seriously studied a century ago, it next century the globally averaged surface tem- ity." has over the last decade emerged as the central perature will rise around 3°C (degrees Celsius), earch international environmental question. Most na- which would produce climates that are unprec- No. tions have adopted the Framework Convention edented during the entire span of human civili- on Climate Change negotiated at the 1992 Rio zation. While warming may seem benign, it has E. O. Earth Summit. Under the Convention, nations major and unpredictable impacts on weather th- agreed to take steps to limit carbon dioxide patterns, ocean currents, sea-level rise, river run- J: (CO₂) and other greenhouse gas (GHG) emis- offs, storm and monsoonal tracks, desertifica- sions before they reach "dangerous" levels. tion, and other geophysical phenomena. Many lign- Having increased its CO₂ emissions at an av- scientists and ecologists view these changes and uring erage growth rate of almost 2 percent annually uncertainties with alarm. vian for about a century, the United States has com- The other half of the calculus is the cost of (2), mitted itself to capping its emissions at 1990 slowing climate change. Even the most dra- levels, and many other high-income countries conian policies (such as a virtual phaseout of Tar- have made similar or even more ambitious fossil fuels) would only slow and not stop cli- ers, proposals (for a review of commitments, see mate change, and significant steps to slow the Daniel M. Bodansky [1995] or International rate of increase of climate change would cost Re- Energy Agency (IEA) [1994]). hundreds of billions of dollars annually using es." The climate-change issue is so controversial today's energy technologies. Given the many Fall primarily because the stakes are so high. If un- economic issues facing humanity, it would re- quire an unusually dire risk and uncommonly lity: statesmanlike behavior for nations to divert 1 cta- * Nordhaus: Department of Economics, Yale Univer- or 2 percent of their national incomes today to sity, 28 Hillhouse Ave., New Haven, CT 06511; Yang: ean reduce conjectural risks that will not occur un- Center for Energy and Environmental Policy Research, pp. MIT, Cambridge, MA 02137. This research was supported til well into the next millennium. by the National Science Foundation and the U.S. Environ- In addition to the grave risks and huge costs, mental Protection Agency. This research has benefited the issue of greenhouse warming is difficult from discussions and comments of Richard Eckaus, because the problem is so complex. It involves William Hogan, Alan Manne, Richard Richels, Herbert Scarf, and two referees. All views and errors of omission a series of poorly understood systems, includ- or commission are the sole responsibility of the authors. ing the carbon cycle, climate reactions, geo- Correspondence can be directed to W.D. Nordhaus. physical, ecological, and biological impacts of 741 742 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 climate changes, economic impacts, along I. Description of the RICE Model with potential adaptations and new technolo- gies, with all of these stretching over a period A. Overview of a century or more. Social and natural sci- entists have made impressive advances in un- This section begins with a succinct descrip- derstanding each of these systems over the last tion of the RICE model; the equations of the quarter century, and numerous efforts are un- model are provided in Appendix A.² The RICE derway today to link together the different model, or Regional Integrated model of Cli- components into an integrated assessment of mate and the Economy, is a regional, dynamic, climate change policies. One of the earliest in- general-equilibrium model of the economy tegrated models was the DICE model, which which integrates economic activity with is a globally aggregated model integrating a the sources, emissions, and consequences of general-equilibrium model of the global econ- greenhouse-gas emissions and climate change. omy with a climate system including emissions, Most existing models of global climate change concentrations, climate change, impacts, and take the vantage point of the Global Com- optimal policy (see Nordhaus, 1992, 1994). moner engaged in determining how nations Other recent integrated models of climate should design sensible strategies to cope with change include Alan S. Manne and Richard future climate change. The RICE model takes Richels (1992), Stephen C. Peck and Thomas a positive point of view by asking how nations J. Teisberg (1992), and Zili Yang (1993). would in practice choose climate-change pol- Globally aggregated models have the short- icies in light of economic trade-offs and na- coming of losing many of the interesting and tional self-interests. Put differently, global C important details of different regions. Perhaps optimization models ask how nations would the central shortcoming, however, is that choose the optimal (or Pareto-efficient) path global models ignore the fact that policy de- for reductions of GHGs. The RICE model S cisions to reduce GHG emissions are taken allows us to calculate not only the efficient primarily at the national level. It is single path (which we designate the cooperative ap- c nations, not the United Nations, that determine proach) but also to compare that path with e energy and environmental policy, so any grand noncooperative approaches. t design to slow global warming must be trans- In the RICE model, the world is divided C lated into national measures. The purpose of into a number of regions. Each is endowed S the present study is to improve the realism of with an initial capital stock, population, and r. integrated assessments by lodging policy mak- technology. Population and technology grow o ing at the more appropriate national level. This exogenously, while capital accumulation is de- involves introducing a number of regions of termined by optimizing the flow of consump- the world and considering different degrees of tion over time. Output is produced by a cooperation among nations. Cobb-Douglas production function in capital, The present paper reports on the results of p labor, and technology. In the long run, capital c the current version of the RICE model.¹ It out- is fully mobile so that the real return on capital lines briefly the philosophy, sketches the mod- eling structure, and describes the major results. b 2 The structural equations of the RICE model are gen- erally the same as those of the aggregated DICE model. 1 An experimental version of the RICE model with il- For a detailed discussion of the derivation of the equations, lustrative data was presented at the MIT Conference on see Nordhaus (1994). The GAMS program for the RICE pl the Environment (see Nordhaus, 1990). The current ver- model is available from the authors upon request. (l sion (called RICE-6.3.2 for purposes of documentation) 3 This study identifies the cooperative solution as the sl incorporates a number of changes, primarily a revision of one that generates an efficient level and distribution of the treatment of non-CO₂ greenhouse gases and improved er emissions. The solutions that might emerge from in- estimates of the economic and emissions data. A major ternational negotiations are a further issue that is not R cause of the long gestation period of this research has been addressed in this study. Issues concerning possible bar- of the difficulty in finding a satisfactory algorithm for solving gaining outcomes are discussed below in Section II.C. er the intertemporal general equilibrium (see below). "Welfare Effects by Region." to VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 743 is equalized across regions. The preference Market policies. The market approach is function of each region is a utility function one in which there are no controls on the which is the sum of discounted utilities of per emissions of greenhouse gases. This has capita consumption times population, where been the approach followed by virtually all :cinct descrip- the pure rate of social time preference (the dis- nations up to now. uations of the count rate on utility) is 3 percent per year in Cooperative policies. The second approach A.² The RICE each region. The utility function is logarithmic is the ideal one in which global environ- model of Cli- in per capita consumption. mental concerns are treated cooperatively mal, dynamic, The major contribution of the integrated ap- through the efficient actions of all nations. the economy proaches like the RICE model is to integrate In this approach, nations agree to reduce activity with the climate-related sectors with the economic CO₂ emissions in a globally efficient way. sequences of model. This part of the model contains a num- This solution is efficient but requires an un- mate change. ber of geophysical relationships that link to- realistically high degree of cooperation. imate change gether the different forces affecting climate Noncooperative policies. In the third ap- Global Com- change, generate the greenhouse-gas emis- proach, individual nations undertake poli- how nations sions, and measure the impacts of climate cies that are in their national self-interests to cope with change. RICE includes region-specific emis- and ignore the spillovers of their actions on model takes sions equations, a global concentrations equa- other nations. In the noncooperative ap- ! how nations tion, a global climate-change equation, and proach, to the extent that nations are small :-change pol- regional climate-damage relationships. En- and the externality is truly global, efforts to offs and na- dogenous emissions are limited to CO2, while reduce CO₂ emissions will be much smaller ently, global other greenhouse gases are treated as exoge- than in the global cooperative solution. This ations would nous. Uncontrolled emissions are a slowly solution is inefficient but realistic. ficient) path declining fraction of gross output-a relation- RICE model ship which is consistent with the observed B. Basic Structure the efficient "decarbonization" in most countries over this ope ? ap- century that is also predicted by more detailed We outline here the major features and in- at with energy models. CO₂ emissions can be con- novations of the RICE model; the equations of trolled by increasing the prices of factors or the model are contained in Appendix A. d is divided outputs that are CO₂ intensive, and we repre- The RICE model divides the global econ- is endowed sent the CO₂-reduction cost schedule paramet- omy into 10 different regions. The first five are ulation, and rically by drawing upon a number of studies 1) the United States, 2) Japan, 3) China, 4) hology grow of the cost of CO₂ reductions. Climate change the European Union, 5) and the former Soviet nation is de- is represented by the realized global mean sur- Union (FSU). Each is treated as a single de- of consump- face temperature, which uses relations based cision maker. The last five regions have dif- uced by a on current climate models. The economic im- ferent numbers of countries, and each is n in capital, pacts of climate change are assumed to be in- treated as multiple decision makers. These five run, capital creasing along with the realized temperature regions are 6) India, 7) Brazil and Indonesia, rn on capital increase. The impacts of climate change are 8) 11 large countries, 9) 38 medium-sized estimated from a number of different studies, countries, and 10) 137 small countries. (Basic but it must be recognized that this is the most data on the major regions are contained in Ap- uncertain part of the model. pendix B.) To reduce the severe computational model are gen- The major economic choices faced by complexity of the solution, we sometimes ag- I DICE model. nations (or the concert of nations) in this ap- gregate regions 6 through 10 into one region I the equations, 1 for the RICE proach are (a) to consume goods and services, as the "rest of the world" or "ROW." equest. (b) to invest in productive capital, and (c) to The goal in creating the different regions is solution as the slow climate change through reducing CO₂ to structure the problem so that the noncoop- distribution of emissions. The new element introduced in the erative equilibrium is equivalent to the full but erge from in- RICE model and not present in other models enormous game with about 200 countries. This ue that is not possible bar- of global warming is the possibility of differ- is done by allocating the smaller countries to Section II.C, ent strategies undertaken by nations. We dis- groups so that within each group the national tinguish three distinct approaches: benefits from slowing climate change are 744 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VOL. 86 NO. 4 roughly equal. We then mimic the free-riding TABLE 1-FUTURE LEVELS OF INCOMES, emissions. 1 temptations of global public goods by dividing DIFFERENT REGIONS United State the benefit function for each region by the number of countries (that is, decision-making Ratio of region's per capita other regions units) within that region. income to that of the have parame United States (US₁₉₉₀ = 1) functional fo An example will clarify the way regions are used. Region 9 contains 38 countries-includ- Region 1990 2100 2200 estimated the on the basis ing Bulgaria and Hungary, which are countries 1) United States 1.00 3.11 4.69 with roughly similar populations and econo- 2) Japan 1.09 4.07 4.83 undertaken b mies. We assume that all the countries in re- 3) China 0.02 0.47 1.55 eling Forum 4) European Union 0.85 2.89 4.27 Estimates ( gion 9 are similar in terms of their sizes, 5) Former Soviet Union 0.14 0.87 2.02 mitigation cost functions, and damage func- ages from cl 6) Rest of the world 0.07 0.84 1.69 tions. Hence, for region 9 the (slightly simpli- stage. There fied) net benefit function to be maximized in Note: These values are the values of per capita GDP gen- mated impact erated by the market solution for the RICE model. The keted sectors the noncooperative case is N(E₉) = B(E₉)/38 GDPs are calculated using market exchange rates. reliable studi C(E₉), where N(E₉) is the net benefits of emissions for region 9, E₉ is emissions in re- for developin gion 9, B(E₉) is the benefit of emissions, 38 pacts in diffe is the number of equal-sized decision makers damage funct tivity differences should largely disappear tical for each in region 9, and C(E₉) is the cost function. over the long run. and that the c Therefore, when the representative country in The assumed ratios of long-run levels of per rameters as 1 region 9 maximizes its net economic welfare capita GDPs to that of the United States are in the noncooperative case, not only will it ig- States. Impac given in Table 1, showing the observed values nore the benefits accruing outside region 9, but culated by tal for 1990 along with projections for 2100 and ferent sectors it will also internalize only 1/38 of the benefits 2200. While highly conjectural, these esti- of the region. This procedure includes in a and so on) in mates are consistent with recent trends in computationally feasible manner all the dif- gregating tho country GDP growth. One interesting feature ferent countries while ensuring that the incen- estimates. (T of this approach is that it gives considerably tives for free-riding are maintained. Nordhaus [19 higher estimates of output and emissions than A major difficulty in constructing the RICE gate do not ( do the conventional global models, such as model has been to estimate the regional pa- major estima those used by governments in the Intergovern- rameters of the different functions.⁴ Gross do- Fankhauser [ mental Panel on Climate Change (IPCC). For mestic products, populations, CO₂ emissions, in Nordhaus I the modeling, each region's income growth is sized that the and capital stocks are taken from a variety of generated through Hicks-neutral technological international sources. Future population growth across countr change, which starts at approximately the ob- estimates are taken from the United Nations jectural. Tabl served rates for 1960-1990. After 1990, projections. The major uncertainty in the eco- sumptions foi growth rates are assumed to decline exponen- nomic projections is long-run levels of per tially in a manner leading to the asymptotic capita output in the different regions. These productivity ratios shown in Table 1. projections are based on the assumption of CO₂ emissions are separated into industrial partial convergence of per capita incomes. emissions (largely from fossil fuels) and those 5 See Andrew That is, we assume that the relative differences from land-use changes and are calibrated to Darius W. Gaskin in regions' per capita incomes decline over 1990 levels. The ratio of CO₂ emissions to out- tional form of the time but do not disappear. The extent of con- model was estima put is assumed initially to decline at different duction in nine fa: vergence is a controversial issue, but to the rates, with each region's decline rate decreas- United States and extent that differences in per capita incomes ing along with the overall rate of technological where i is region i are primarily based on differences in the extent change by region. Here again, asymptotic sions, b₁., and b₂ of adoption of available technologies, produc- control rate or fra- CO2-output ratios are assumed to converge market path, Y,(t) considerably but not completely in the future. 1 is the time peri The costs of reducing emissions by region exponents (b₂) are 4 A detailed list of sources and data are available from are estimated separately on the basis of the ex- the intercepts (b₁., the authors. the different cou isting studies of the cost of reduction of CO₂ above. ER 1996 VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 745 emissions. Most studies are based on the The climate-change policies are character- United States and Europe, and estimates for ized by "control rates" and "carbon taxes." other regions have low levels of reliability. We Control rates are simply the percentage reduc- er capita have parametrized the cost function using the tions in CO₂ emissions relative to a baseline of the 1990 = 1) functional form from earlier studies but have or uncontrolled path. Carbon taxes represent 2200 estimated the intercepts of the cost functions the marginal cost of reducing CO₂ emissions. on the basis of the international comparisons A carbon tax would equal the price of a 4.69 undertaken by the OECD and by Energy Mod- carbon-emissions permit if there were tradable 4.83 1.55 eling Forum 13.⁵ permits, and the prices of such permits in dif- 4.27 Estimates of the economic impacts or dam- ferent countries would obviously be equalized 2.02 ages from climate change are sparse at this (at market exchange rates) if permits were 1.69 stage. There are numerous studies of the esti- freely tradable. In the market solution, carbon GDP gen- mated impact of climate change on the mar- taxes are zero. In the cooperative solution, keted sectors for the United States, but few emissions are curtailed in a cost-effective odel. The S. reliable studies for the nonmarket sectors or manner. The model does not deal explicitly for developing countries. To estimate the im- with mechanisms by which winners might pacts in different regions, we assume that the compensate losers, although we discuss some damage function from climate change is iden- of the issues below. sappear tical for each industry across different regions, and that the cost functions have the same pa- C. Algorithm to Calculate is of per rameters as those estimated for the United General Equilibrium ates are States. Impacts in different regions are cal- 1 values culated by taking the estimated shares of dif- The RICE model presents a radically differ- 100 and ferent sectors (agriculture, coastal activities, ent philosophy for estimating strategies to se esti- and so on) in national output and then ag- cope with global warming from global- ends in gregating those up to obtain overall national optimization models used in many integrated feature estimates. (This approach is described in assessments. The baseline calculation is cali- ider21 Nordhaus [1994].) The results in the aggre- brated to a market equilibrium of the world ons gate do not differ markedly from the other economy with all the differences in popula- suc. major estimates (see particularly Samuel tions, technologies, and incomes-the world govern- Fankhauser [1993] and the survey of experts is taken as it is for the purpose of the baseline 'C). For in Nordhaus [1994]), but it must be empha- calibration. We then calculate different strat- rowth is sized that the distribution of climate impacts egies for global warming conditional on the ological across countries is at this stage highly con- existing distribution of capital, labor, and tech- the ob- jectural. Table 2 shows the major inputs as- nology. The strategies include doing nothing r 1990, sumptions for the different regions. (the market solution), finding an efficient xponen- solution given the existing distribution of in- vmptotic come (the cooperative solution), and finding the solution in which nations select policies to ndustrial maximize national preferences alone (the non- nd those 5 See Andrew Dean and Peter Hoeller (1992) and Darius W. Gaskins and John P. Weyant (1993). The func- cooperative or nationalistic equilibrium). This rated to tional form of the mitigation-cost function in the DICE public-choice approach is in sharp contrast to is to out- model was estimated from studies of the cost of CO₂ re- many of the debates on climate change today; different duction in nine families of models primarily based on the in these, the distributional issues of who shall decreas- United States and takes the form C,(t) = b₁.ᵢµ,(t)b₂Y₁(t), pay to slow climate change rise to the top of hological where i is region i, Cᵢ(t) = the cost of reducing CO₂ emis- sions, b₁, and b₂ are parameters, µᵢ(t) is the emission- the agenda. ymptotic control rate or fractional reduction in emissions from the We now describe the algorithm for finding converge market path, Y,(t) is region i's gross regional product, and the cooperative solution in the RICE model. e future. 1 is the time period. The RICE model assumes that the The technique we employ originates with y region exponents (b₂) are the same across countries and calibrates T. Negishi (1960), was discussed briefly in the intercepts (b₁.ᵢ) to estimates of the cost functions from of the ex- the different countries or regional models mentioned Nordhaus (1990) in the context of global 1 of CO₂ above. warming, and has been used in similar models 746 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VOL. 86 NO. TABLE 2-MAJOR INPUT PARAMETERS FOR THE RICE MODEL an arbitrar would gen Climate CO₂ emissions, 1990 Per capita which imp) Cost damage Population output CO₂ ratio, live within Region intercept" interceptb Land-useᶜ Industrialᵈ 2100° (2100)' 2100' variables ai United States 0.07 0.01102 0.010 1.360 0.294 68.8 0.1190 welfare-we Japan 0.05 0.01174 0.000 0.292 0.125 89.1 0.0630 are equal : China 0.15 0.01523 0.136 0.669 1.656 9.9 0.5120 European Union 0.05 0.01174 0.100 0.872 0.427 63.0 0.0740 straints is 1 Former Soviet Union 0.15 0.00857 0.000 1.066 0.366 18.9 0.3220 Hence, t Rest of world 0.10 0.02093 1.730 1.700 6.738 18.1 1.1850 the welfare variables 4 a The intercept of cost function equals the fraction of annual output required to reduce net CO₂ emissions to 0. of consum b The intercept of climate-damage function equals the reduction in annual net output from an increase of 2.5°C in global mean temperature. c Emissions are measured in billions of tons carbon per year. Land-use emissions are primarily from deforestation. (3) φ'(7 d Emissions are measured in billions of tons carbon per year. Industrial uses primarily from burning fossil fuels. e Population is in billions of people. 1 Gross domestic product (GDP) is measured at 1990 market exchange rates in thousands of 1990 U.S. dollars. & The ratio is of industrial CO₂ emissions to GDP (tons of carbon per $1000 of output in 1990 U.S. dollars). Combinin: each of the isfied and by Manne and Thomas Rutherford (see, par- optimizing a global social welfare function of change in ticularly, 1994). The theoretical basis for the the form: the Negisl algorithm is a theorem of Negishi which relies timized 01 on the second theorem of welfare economics. N generated Negishi suggested and proved that under cer- (1) W = Σ c¹(2), equi-librit i=1 tain conditions a competitive equilibrium can dowments be found by maximizing a social welfare func- c'(t), c'(T)] equilibriu tion of N agents in which the welfare weight solution.' of each of the agents is adjusted to satisfy the where W is the value of the global social wel- Unsatis agent's budget constraint. We will call this fare function and Φ¹ are the welfare or Negishi to the fo equilibrium the Negishi solution. weights for country i, i = 1, N. The U' are Negishi S( What are the appropriate welfare weights? the preference functions for the different coun- pure Negi In our calibration, we adopt the realistic ap- tries, and the c'(t) are the consumption bun- tremely 1. proach by taking the welfare weights that re- dles of the countries. (this is a flect the actual economic outcome across The relevant excess demand is found in the optimized regions. We do this not as a brief for the ex- intertemporal budget constraint of each region. realistic, isting international distribution of resources To find the competitive equilibrium, we add a to impose and income but because it is the starting point constraint to the problem that requires each re- on debt ai for analyzing potential improvements in eco- gion to satisfy its intertemporal budget con- foreign in nomic welfare that would arise from policies straint, which is represented by terminal net limits. TI that are imposed on the actual world economy. foreign assets, NFA'(T), T being the last GDP ratio Hence, the weights are ones such that the ex- period: assets to ( cess demands in all markets are zero at the account C given welfare weights and prices.⁶ More pre- (2) NFA'(T) = 0, i = 1, .. , N. pendix A cisely, the algorithmic procedure is the follow- based on ing. We first solve the RICE model by Next, define φ'(T) as the dual variable of which in economic terms is the marginal utility of consumption or income in the last period. Given condition (2), φ'(T) is 7 The dit a function of the welfare weights and we can 6 A brief but illuminating discussion of the Negishi ap- flows is not proach is in contained in Andreu Mas-Colell et al. (1995 write these functions as φ'(T) = G'(Φ¹, Φ², the same is pp. 630-31). Φ), i Il 1, N. Without condition (2), employed. !BER 1996 VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 747 an arbitrary set of social welfare weights virtually no difference for the results of the would generate a set of nonzero NFA'(T), analysis below. which implies that at least one region does not Given these constraints on international CO₂ ratio, live within its budget. However, when the dual 2100 capital flows, our algorithm will not produce variables are equalized across all countries, the the necessary complete price equalization for 0.1190 welfare-weighted marginal utilities of income carbon-trading permits, which are assumed to 0.0630 0.5120 are equal and the intertemporal budget con- be fully tradable and reach price. To ensure 0.0740 straints is therefore satisfied. price equalization for carbon-emission rights, 0.3220 Hence, the algorithm works by searching for we adjust. the Negishi weights across regions 1.1850 the welfare weights, as a function of the dual for every period. We call this new algorithm variables φᵢ(T), so that the marginal utilities to 0. the time-dependent Negishi solution. It differs of 2.5°C in of consumption are equalized: from the pure Negishi solution because it in- corporates the constraints on capital flows so restation. (3) = Φ) = that the regional budget constraints are binding 1 fuels. for every period. As a result, carbon-emissions llars. for all i = 1, N. permits have equal prices in all regions in each s). time period (at market exchange rates). Under Combining (1), (2), and (3), we know that this revised algorithm, we seek the time- each of the country budget constraints is sat- dependent Negishi weights, Φ'(t). To find isfied and that no region can gain from a. these, we first solve the model with an arbi- inction of change in the resulting allocation. Hence, by trary set of welfare weights while continuing the Negishi theorem, we know that this op- to impose (2). Following the Negishi theorem, timized outcome using the welfare weights we then reset the welfare weights for all coun- generated in (3) represents a competitive tries and time periods according to the follow- equi-librium consistent with the initial en- ing formula: dowments, technology, and preferences. The equilibrium thus found is the "pure Negishi 1 solution." sci Unsatisfactory aspects of the solution led (4) or to the following refinements of the pure "he Ui are Negishi solution. The major problem with the φ'(t) rent coun- pure Negishi solution was that it generated ex- tion bun- tremely large capital flows among regions This equation sets the welfare weights equal (this is a common feature in intertemporally to the inverse of the marginal utilities of con- und in the optimized models).⁷ Because these are un- sumption. The search algorithm based on (4) ch region. realistic, we took one further step which was very quickly converges to a solution that sat- we add a to impose certain flow and stock constraints isfies (2) and (3). We have conducted a es each re- on debt and current accounts to ensure that net number of experiments and have found no in- dget con- foreign investment does not exceed certain dication of multiple equilibria. minal net limits. These limitations limited the export- What is the underlying economic rationale ; the last GDP ratio to 1, limited the ratio of net foreign for this algorithm? The solution represents a assets to output to 0.1, and limited the current competitive equilibrium under the assumption account deficit to GDP ratio to 0.1 (see Ap- that the preferences or technological con- N. pendix A for details). These constraints were straints limit the international flows of capital. based on observed limitations, but they made For example, there may be strong home- ariable of country preferences in portfolios because of ms is the limitations of the marketability of human cap- income in ital. The limitation of this approach is that φ'(T) is 7 The difficulty raised by unrealistically high capital to the extent that the constraints on capital nd we can flows is not related to the use of the Negishi technique; flows have nonmarket-clearing elements due (Φ¹, Φ², the same issue would occur if fixed-point methods were to rationing, the excess demands will not be dition (2), employed. zero and we may depart from the market 748 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VC equilibrium (which would in any case be dif- regions by optimizing for each region holding T ficult to compute). the control rates and resulting emissions, con- Once we have obtained a competitive equi- centrations, and impacts in other regions from librium, we then perturb various elements, the previous iteration fixed. We continue to 1. such as the climate parameters or the cost cycle through this sequence until the set of functions, and resolve the maximization in control rates are unchanged given the set of (1). We do this holding the welfare weights noncooperative strategies of other countries, 2. constant across runs. This resolves the index which is then the Nash equilibrium. The out- number problem of changing prices by calcu- come matches well the theoretical predictions lating the welfare changes at the market wel- and is in our simulations invariant to initial fare weights.⁸ conditions, which suggests that the Nash equi- librium is unique. 3. D. Finding the Noncooperative Equilibrium How reasonable is this solution concept? While the pure Nash equilibrium is a sensi- The algorithm just described provides the ble assumption for small countries like Chad, solutions for both the market and the cooper- whose global warming policies will hardly ative equilibrium. A different approach is nec- make the front pages, it may lack realism for essary to find the noncooperative equilibrium. large or influential countries. Large coun- The noncooperative or nationalistic equilib- tries like China and influential countries like rium exists as the equilibrium of the strategies the United States would probably want to of the different countries. We hence need an take into account the effect of their policies assumption about strategies and a method of on other countries' policies. The ambivalent finding the equilibrium. policy on global warming by the United str: As for strategies, we assume that each na- States over the last decade has undoubtedly eff tion determines its policies by maximizing its strengthened the hand of those in other coun- wit domestic intertemporal utility function assum- tries who want to do little. An alternative app ing that other nations' strategies are unaffected approach would be for countries to posit glc by its policies. The noncooperative strategies conjectural variations or reactions of other har are hence dynamic, full-information, Nash countries to their policies. For example, the stri strategies, and we are seeking the Nash equi- United States might assume that Japan or Eu- be librium. Technically, our solution is a Nash rope would be a follower in terms of carbon- mc equilibrium in a finite game with perfect in- tax policies or tradable emissions policies. RI formation, and it is therefore time consistent. Another possibility would be to model coa- fid Such games have pure strategy Nash equilibria litions of different countries. We have not col which can be calculated through backward in- explored these alternative solution strategies jor duction, which is essentially what our algo- in the present paper. Once we admit nonzero lon rithm does (for a discussion, see Mas-Colell conjectural variations, we are in a deep dai et al. [1995 Chapter 9]). thicket and the possibilities become unlim- More precisely, we assume that each nation ited. Future research will examine the pos- A. sets its own control rate over time [µ' = sibility of coalitions of countries. {µ'(1), µ'(2), µ'(3), µ'(t), µ'(T)}; i = 1, , N] so as to maximize its national E. The Economic and Environmental Impact and objective function taking the control rates of of Alternative Strategies Fig the other regions {µ¹, , µⁱ⁺ of 1 µⁿ} as given. Beginning with an initial set of Using the algorithms just described, we will sta control rates, we iterate through the different analyze the three different strategies as de- em scribed in Table 3: market, cooperative, and do noncooperative. In addition, for reference we as sometimes compare the results of the RICE 8The RICE model runs on the GAMS software (see Anthony Brooke et al., 1988). The full model including model to those of its parent, the DICE model, searching for welfare weights takes approximately 6 hours which is essentially a one-region efficient or 9 on a 486-66 processor. cooperative solution. in t ER VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 749 holding TABLE 3-ALTERNATIVE SOLUTION CONCEPTS FOR THE regional outputs are shown in Figure 1; these IS, con- RICE MODEL indicate that the projected relative sizes of as from the Chinese and ROW economies grow inue to 1. Market RICE: This strategy assumes that there set of is no correction for the climate-change sharply over the next century. The output externality and that there is therefore no growth in the RICE model is significantly set of abatement of CO2 emissions. larger than that in many projections prepared untries, 2. Cooperative RICE: In this strategy, countries by international study groups, most of which ne out- undertake policies that reduce greenhouse-gas envision a stability of current relative in- lictions emissions efficiently. The reduction of CO₂ come differentials rather than the projected initial emissions is efficient across countries and partial convergence in the RICE model. Note h equi- across time. as well that we use market exchange rates 3. Noncooperative RICE: This strategic concept because we will want to find the equilibrium ncept? assumes that each country sets its CO₂ in which the prices of internationally-traded sensi- emissions controls to maximize its own Chad, economic welfare assuming that other carbon-emissions permits are equalized. countries' control strategies are invariant to a Emissions are also considerably higher in hardly country's policies. RICE than in the many other projections. For sm for example, CO₂ emissions in the RICE model coun- reach 38 billion tons of carbon by the year es like 2100 in the market or uncontrolled run. This ant to II. Results compares with an estimated 21 billion tons in olicies the DICE model and a range of 5 to 35 billion valent We now report the results of the policies and tons in the IPCC projections (see T. M. L. United strategies described above. As in all modeling Wigley [1994] for a description). CO₂ emis- btedly efforts of this kind, they should be interpreted sions grow substantially faster in the RICE coun- with caution as this study is the first empirical model partially because of the projected rapid native application of noncooperative game theory to growth in output and partially because of the T global environmental policy. On the other rising output share of regions with high o hand, the major results concerning the level of emission-output ratios. ie, the stringency of climate-change policies have Figure 2 shows the resulting CO2 emissions or Eu- been relatively stable over a wide variety of under the different solution concepts and also arbon- models and alternative specifications of the compares estimates from this study with the licies. RICE model, so we have considerable con- earlier DICE model. Model estimates (not el coa- fidence in these estimates (conditional, of shown) indicate that the share of CO₂ emis- ve not course, on the assumptions underlying the ma- sions will rise sharply in China, region S1 itegies jor components, such as those concerning the (India), region S3 (middle-sized developing onzero long-run growth projections, the costs and countries like Thailand), and region S4 (smaller deep damages, and the discount rate). developing countries). These four regions ac- unlim- counted for about one third of CO2 emissions e pos- A. Output, Emissions, and Climate Change in 1990 but are projected in the market runs of the RICE model to comprise three quarters of The projections for the major economic emissions by 2100. mpact and environmental variables are shown as CO₂ concentrations are shown in Figure 3. Figures 1 through 4. One important outcome Given the higher emissions rates in the RICE of this study is that the RICE model has sub- model, its concentrations rise more rapidly than ve will stantially higher projected world output and in the DICE model. It is useful to examine the as de- emissions by the end of the next century than date of doubling of CO₂ concentrations relative and do many other integrated assessments, such ce we as the earlier DICE model.⁹ Projections for RICE nodel, emmental Panel on Climate Change (IPCC) (1990), the re- sults in Nordhaus (1994), and preliminary results of the sur- ent or This statement is based on a comparison of the results vey of models by the Energy Modeling Forum 14 directed in the RICE model with the projections of the Intergov- by John Weyant, Energy Modeling Forum (EMF) (1995). 750 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VOL. 86 N( 100.0 40 Rest of World 35 Europe 30 Output (trillons of 1990 U.S. dollars) * Billion tons of carbon per year 25 - 10.0 USA X 20 FSU China 15 Japan 10 1.0 1990 2010 2030 2050 2070 2090 5 FIGURE 1. REGIONAL OUTPUTS: COOPERATIVE SCENARIO to preindustrial concentrations; that benchmark strategies reduce warming by considerably less is taken to be 1,200 billion tons of atmospheric (a reduction of 0.086°C in 2100), both com- son for CO₂ concentrations (or 565 parts per million of pared to the market strategy. One reason that between CO₂). The doubling date is 2100 in the (coop- the difference in the temperature increase be- that the i erative) DICE model, 2070 in the cooperative tween the cooperative and the market runs is so economic RICE model, and 2065 in both the market and small is because of the long time lag between small red the noncooperative model. The doubling time changes in emissions and temperature increases for the CO₂ equivalent of all greenhouse gases (the difference between the runs grows over is slightly earlier than those for CO₂ alone. time as the lags in the emissions-concentrations- The projected increase in global mean tem- temperature relationship plays out). 11 Addi- We ne perature over the 1990-2100 period is shown tionally, the difference is small because of the the diffe in Figure 4.¹⁰ The estimated temperature in- nonlinear relationship between CO₂ concen- nations ( crease from the mid-nineteenth century to 2100 trations and temperature. 12 But the major rea- ative). T is estimated to be 3.06°C in the market run. The the conti cooperative strategy lowers global temperature carbon ta by 0.22°C in 2100, whereas the noncooperative II The projected temperature difference between the coop- preted as erative and market runs is 0.41°C in 2200 whereas that between whether the noncooperative and market runs is only 0.12°C in 2200. 12 More recent estimates of global warming show con- through t siderably less near-term warming than earlier estimates The m 10 The climate model used in the RICE model is a cal- (compare the current RICE with the 1992 DICE model). through ibrated version of the two-equation Schneider-Thompson Recent evidence suggests that the cooling effects of sul- that the : model with an equilibrium temperature sensitivity coefficient fates derived primarily from fossil-fuel emissions will of 3°C for a doubling of CO₂ concentrations. The derivation lower global mean temperature increases until the end of nificantly of the climate model is discussed in Nordhaus (1994). the next century. than doe TE 1996 VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 751 40 Market solution 35 Noncooperative e equilibrium 30 Billion tons of carbon per year 25 Cooperative equilibrium 20 FSU DICE model 15 10 90 5 1990 2010 2030 2050 2070 2090 FIGURE 2. CO₂ EMISSIONS: DIFFERENT APPROACHES siderably less 1, both com- son for the small decrease in temperature straightforward: when countries free-ride on : reason that between the market and cooperative runs is the climate-change policies of other countries, increase be- that the high cost of control means that the then they cut back their own efforts substan- ket runs is so economically efficient strategy is for only a tially. Begin with the emissions control rates, lag between small reduction in CO₂ emissions. shown in Figure 5. The global average rate of ure increases control of CO₂ is around 10 percent in the co- grows over B. Policy Variables operative solution. This varies by region, with oncentrations- relatively high controls in China and the former ut). 11 Addi- We next examine the policy variables for Soviet Union; for these regions, we estimate the ecause of the the different degrees of cooperation among marginal costs of control to be relatively low. CO₂ concen- nations (market, noncooperative, and cooper- For the efficient case, the lowest control rates e major rea- ative). The results are shown in terms of both are in Japan and the European Union, which the control rates for CO₂ emissions and the are already relatively energy efficient and carbon taxes. Carbon taxes should be inter- where the marginal costs of controls are con- :tween the coop- preted as the marginal cost of control of CO₂ reas that between sequently relatively high. According to the data 1.12°C in 2200. whether these are efficiently implemented used in the RICE model, the efficient control ning show con- through taxes, regulations, or tradable permits. rates for 2000 range from 17 percent in China arlier estimates The major results are shown in Figures 5 to about 7 percent in Japan. The United States 2 DICE model). through 9. The central finding of this study is is in the middle of the pack, with an efficient or effects of sul- emissions will that the noncooperative policies produce sig- control rate of slightly below 9 percent. The until the end of nificantly lower control rates and carbon taxes control rates rise over time as the marginal than does global cooperation. The reason is damages from CO₂ emissions rise. (Note that 752 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VO 1800 Market solution 1600 Noncooperative equilibrium Concentrations (billions tons C) 1400 1200 Cooperative 1000 equilibrium Change in temperature (degrees C) 1 DICE model 1 800 0 600 1990 2010 2030 2050 2070 2090 FIGURE 3. CO₂ CONCENTRATIONS 0. these relative control rates would be roughly with the average of 9.7 percent in the cooper- proportional to those shown here if the overall ative case. The reason for the lower control rate level of controls were raised or lowered.) is completely intuitive: it results from the free- One immediate conclusion that comes from riding wherein each nation ignores the impacts under this result is that current approaches to com- of its CO₂ emissions on the welfare of other erativ bating global warming make no sense from the nations (as well, of course, as assuming that the C point of view of pure economic efficiency. The other nations' efforts are unaffected by its own the g current Framework Convention calls for major self-interested behavior). The size of the free- slight emissions reductions in the OECD region with riding effect is the major new result here. cause no immediate reductions in the developing The second interesting conclusion in the The f countries-this being exactly the opposite of noncooperative approach is the distribution of case i the efficient solution. The only potential ration- control rates. This model predicts that the larg- $5.94 ale for the Framework Convention is that it puts est (albeit small) efforts will be taken by the out, a a very high weight on equity (by relieving poor largest regions-particularly by the United U.S. ( countries of obligations to reduce emissions) States and the European Union. This predic- The and rules out the possibility of side payments tion seems quite on the mark. It also correctly higher (say through allocation of emissions permits). suggests that developing countries, particu- policie The control rates in the noncooperative solu- larly small and poor countries such as Benin weigh tion are markedly lower (not shown but avail- and Kyrgyzstan, will not be in the forefront of operat able from the authors). There are two major global-warming politics. the no findings here. First, the aggregate global emis- Figures 6 through 7 show the results for es- distrib sions control rate for the noncooperative equi- timated carbon taxes. Looking first at Figure erative librium is in 2000 only 2.3 percent as compared 6, we can compare the aggregate carbon taxes nonco IS VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 753 3.5 Market solution 3.0 * 2.5 Change in temperature (degrees C) Noncooperative 2.0 equilibrium DICE model 1.5 Cooperative equilibrium 1.0 0.5 0.0 1990 2010 2030 2050 2070 2090 FIGURE 4. TEMPERATURE CHANGE oper- 1 rate free- pacts under different strategies. Note that the coop- to have significantly more (but not very) strin- other erative RICE model looks quite similar to gent controls as compared to small countries. that the older DICE model (which also found The noncooperative carbon taxes are highest own the global optimum). The carbon tax starts in the European Union ($0.86 per ton in 2000) free- slightly higher and grows more rapidly be- and the United States ($0.65 per ton in 2000). cause of the steeper trajectory for emissions. The difference reflects the slightly larger out- 1 the The first-period carbon tax in the cooperative put in the European Union. For smaller coun- on of case is $6.19 per ton carbon in 2000 versus tries, the tax rates are much smaller: 10 cents larg- $5.94 in the DICE model. (Here and through- per ton in India, and only 1 cent per ton in the y the out, all dollar figures refer to prices in 1990 S4 group of countries. nited U.S. dollars at 1990 market exchange rates.) It seems appropriate to conclude that outside redic- The cooperative tax rates are significantly the United States, Europe, and Japan, the ra- rectly higher than the noncooperative or nationalistic tional noncooperative strategy would be sim- rticu- policies for all regions and periods. The ply to ignore global warming at the present Benin weighted average carbon tax for the nonco- time. Even by the end of the 21st century, no ont of operative policy is 24 cents per ton carbon for country acting in a noncooperative framework the noncooperation equilibrium in 2000. The would have carbon taxes above $2 per ton C. or es- distribution of carbon taxes for the noncoop- If we define the "cooperation ratio" as the Figure erative policy is shown in Figure 7. For the ratio of the noncooperative carbon tax to the taxes noncooperative strategies, large countries tend cooperative carbon tax, we can calculate that 754 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VOI 0.25 China 0.20 FSU 0.15 Control rate by region ROW 0.10 USA Japan Europe Carbon tax (1990 dollars ner ton) 0.05 0.00 1990 2010 2030 2050 2070 2090 FIGURE 5. CO₂ CONTROL RATES: COOPERATIVE SCENARIO this ratio ranges from essentially zero in the than those in the global cooperative strategies, smallest countries to between 10 and 15 per- some have expressed surprise that they are sma cent for the United States and Europe. not even lower. The reason is that there are a emis What happens to the cooperation ratio over few countries or regions (notably the United and time? According to our calculations, the de- States, China, Japan, and Europe) which are gain gree of cooperation is expected to fall in the large enough so that it is their own self interest noncooperative solution. Cooperation in the to reduce CO₂ emissions even ignoring the Nash equilibrium decreases as the extent of benefits to other countries. Were China to inequality of country income falls. Hence, the break up, were Europe to make decisions on a W extent of cooperation is calculated to decline national level, or were the Republican Revo- regi- slightly over the next four decades as the share lution in the United States to devolve environ- latec of the United States, Japan, and Europe declines mental decisions to the states, the predicted cons and the distribution of economic sizes of nations degree of cooperation would be even lower. disc becomes more equal. Greater equality leads to There are a few other intriguing details of fuse smaller incentives to be a good global citizen. the runs worth noting. China is definitely a key eren For small countries (with GDPs of under player and exhibits a different pattern. Figure equa $20 billion) the noncooperative optimal con- 5 shows that China has the highest cooperative in th trol rates and carbon taxes are minuscule, control rates of all the regions-this reflecting cific $0.01 per ton carbon versus $5.98 in the global the relatively high CO₂ emissions per unit out- year cooperative case. While the taxes in the non- put (see Table 2). But countries which are these cooperative strategies are significantly lower hardly players today (India, China, and the whic 190 VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 755 30 Cooperative equilibrium 25 Carbon tax (1990 U.S. dollars per ton) 20 15 DICE model 10 Noncooperative Market equilibrium 5 solution 0 1990 2010 2030 2050 2070 2090 FIGURE 6. AVERAGE CARBON TAXES ies, are smaller developing countries) dominate CO₂ is assigned its optimal policy without any e a emissions by the middle of the next century side payments from other countries. This is ted and will have to behave cooperatively if the equivalent to each country receiving in the co- are gains from cooperation are to be realized. operative equilibrium a quota of tradable emis- rest sions permits equal to the quantity of its own the C. Welfare Effects by Region emissions. to The resulting impacts upon economic wel- na What are the overall economic effects by fare are shown in Table 4. Note first that the vo- region? The gain to cooperation is calcu- overall results from the cooperative RICE so- on- lated as the present value of the change in lution are quite close to those of the original ted consumption valued using the region-specific DICE model. The former is about one quarter discount rates on consumption (not to be con- higher because of the higher growth rates in of fused with the pure rates of social time pref- the RICE model. By contrast, the noncooper- tey erence, or discount rates on utility, which are ative, six-region RICE model shows extremely are equal across regions). The discount rates slim net benefits-only $43 billion in dis- ve in this calculation are region and time spe- counted benefits as opposed to $344 billion for ng cific, and they average about 4½ percent per the cooperative RICE or $271 for the coop- ut- year (in real terms) over the next century. In erative DICE model. are these runs, there are no international transfers, Figure 8 shows the gains to different regions the which essentially means that each country for the cooperative and noncooperative cases. 756 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 1.5 Europe USA S M Carbon tax (1990 U.S. dollars per ton C) N 1.0 C D N Japan FSU S} 0.5 S) in China is p tl St a: a: 0.0 p 1990 2010 2030 2050 W re FIGURE 7. NONCOOPERATIVE CARBON TAXES d S M Table 4 and Figure 8 present a number of sur- benefits in the cooperative strategy because it e) prises in the regional results. The noncooper- is required to undertake significant mitigation CI ative solution produces positive net benefits efforts and has few benefits because of its u relative to the market solution for all regions. northerly location. By contrast, the ROW re- u This result is expected because the noncoop- gion reaps major net benefits from the coop- a erative policies improve welfare while the ex- erative solution because the mitigation efforts to ternal interactions among countries are ones are undertaken primarily in the high-income a that are beneficial relative to the market case. countries early in time while the major benefits The net benefits in the noncooperative case are in terms of damages avoided accrue to the de- sl relatively uniform across the different regions, veloping countries in several decades. with most of the positive effects coming from These results indicate that the cooperative ci the reductions in damage from climate change. solution-one in which nations are allocated The major surprise in these results is the lop- emissions equals to their efficient emissions- ai sided benefits from the cooperative strategy. might well not emerge as the outcome of a The United States actually loses in the coop- bargaining process in which nations will only erative solution relative to the noncooperative sign on to an agreement that improves their equilibrium. The reason is that, with its rela- economic welfare. Of course, the pattern of net di tively large emissions, the United States would gains can in principle be altered through dif- di be slated to incur major costs today, while its ferent schemes for allocating emissions rights benefits would be relatively small given its de- to countries (that is, by adding side payments W clining share of the world economy. Similarly, to the program analyzed here); the gains and re the former Soviet Union has quite modest net losses could be made much more equal over p El VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 757 TABLE 4-NET BENEFITS OF DIFFERENT STRATEGIES BY REGION RELATIVE TO THE MARKET EQUILIBRIUM (BILLIONS OF 1990 U.S. DOLLARS, DISCOUNTED TO 1990) Net benefits by region European Former Soviet Rest of Strategy United States Japan China Union Union world Total Market 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Noncooperative 2.9 3.6 8.7 7.9 2.7 16.5 42.5 Cooperative 0.8 46.3 39.4 28.5 4.1 224.8 343.8 DICE (cooperative)' na na na na na na 271.0 Note: Each entry indicates the net benefits for a region relative to the market or uncontrolled strategy. NA is not available. a From the aggregate DICE model in Nordhaus (1994). space and time through different allocations or market solution through different time periods. side payments. Determining possible bargain- For example, it shows that the United States ing outcomes is, however, a difficult empirical would have a cumulative discounted con- issue that is outside the scope of the present sumption loss from cooperation relative to the paper and is the subject of current research by market of $12 billion through 2050. The cal- the authors. What this study examines is the culation indicates that a cooperative global- set of national emissions that is consistent with warming accord would reduce the cumulative an efficient allocations of emissions over space discounted consumption of all countries ex- and time. The interesting new result of this cept Japan through 2050. The ROW region paper is that a scheme with no side payments suffers major losses, approaching a total of will reduce the standards of living of all major $100 billion by mid-century. Moreover, as can regions for at least half a century and will re- be seen by adding the numbers for the different duce the discounted net welfare of the United regions together, there is still a negative effect States when all time periods are considered. on cumulative global consumption by the mid- Moreover, it is interesting to note that all the dle of the next century. use it emission-rights allocations proposals that are On a longer time scale (not shown), the gation currently under consideration are even more ROW breaks even by the end of the next cen- of its unfavorable to the United States than the one tury and is the major beneficiary after that W re- underlying the cooperative equilibrium and point. The United States and the former Soviet coop- are therefore even less likely to be acceptable Union experience a reduction in discounted efforts to high-income countries than the program ex- cumulative consumption through the end of icome amined here. the next century. All the curves are heading up :nefits What is the time profile of benefits? Figure 9 at the end of the period, and the discounted he de- shows the time paths of discounted cumulative cumulative totals over the 250-year estimation consumption in different regions. More pre- period, shown in Table 4, are positive for all rative cisely, the numbers are the sum of the consump- regions and quite large for the ROW region. cated tion differences between the cooperative strategy The estimates of the regional costs and ons- and the market strategy from the beginning of benefits in the RICE model are sensitive to pa- of a the period (1990) until the date shown on the rameters of the mitigation-cost and climate- only horizontal axis. For each region, the consump- damage functions, but the major determinant their tion figures are discounted back to 1990 and the of the patterns is initial emissions and growth of net discount rate is the region-specific and variable of output, which are considerably more secure 7 dif- discount rate on consumption. than the cost and damage estimates. The basic rights This figure shows the problem of global dilemma is clear: the long period between nents warming in a nutshell. It indicates how each emissions reductions and reduced climate dam- and region would experience the economic im- age means that countries must be extraordi- over pacts of a cooperative strategy relative to the narily farsighted. In addition, the pattern of 758 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VOL. 350 300 Gains (billions of 1990 U.S. dollars) 250 200 150 100 50 Difference in consumption (bill of dollars) 0 USA JPN CHN EEC FSU ROW World Cooperative strategy Noncooperative strategy FIGURE 8. GAINS FROM CLIMATE-CHANGE POLICY: COOPERATIVE AND NONCOOPERATIVE POLICIES (TOTAL GAINS DISCOUNTED TO 1990) gains and losses, with the major long-run gains uncertainties are often critical to determining coming to developing countries while the net policies, formal techniques for determining benefits to the United States and the former the uncertainty of future trajectories or of im- 199 Soviet Union are minimal, is a most surprising pacts have been rarely applied to major policy equa and troubling finding. issues.¹⁴ stan- A full-scale analysis of the uncertainties as- pres D. Sensitivity Analysis sociated with the RICE model-including un- resp certainty about model structure as well as mod To understand the full range of outcomes about individual parameters-is beyond the of th and policy responses to the threat of global scope of the current article. Many of the cen- A), warming, we must assess the fact that many of tral uncertainties have been examined in the it fr the underlying processes are imperfectly un- context of the DICE model (see Nordhaus, subj derstood. Social scientists have developed a vatio variety of tools to incorporate uncertainty into quantitative modeling, and these can help put bounds on potential future outcomes.¹³ Although 14 One notable and controversial example of the sys- 15 tematic application of statistical techniques is the Ras- mapp mussen report (Nuclear Regulatory Commission, 1975), doger which estimated the risk of accidents of different levels of 13 See M. Granger Morgan and Max Henrion (1990) and I severity in commercial nuclear power plants. An exem- for a recent survey of tools for the analysis of uncertainty certai plary study used probabilistic assessments for ozone de- in quantitative risk and policy analysis. the pletion (National Academy of Sciences, 1979). 50th 'BE VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 759 10 China Japan Europe 0 Difference in consumption (bill of dollars) -10 USA FSU -20 -30 -40 Rest of World -50 1990 2010 2030 2050 FIGURE 9. CUMULATIVE DISCOUNTED CONSUMPTION: COOPERATIVE VERSUS MARKET STRATEGY (TOTAL GAINS FOR CONSUMPTION THROUGH THE GIVEN DATE, DISCOUNTED BACK TO 1990) nining nining of im- 1994 Chapters 6-8), and those results apply in Nordhaus (1994 Table 6.1), and the reader policy equally well to the RICE model. To under- is referred to that reference for a full discussion. stand the extent of sensitivity of the model we Figure 10 shows the results of the sensitivity ies as- present here a limited sensitivity analysis with analysis. That figure shows the sensitivity of respect to the important parameters of the 1g un- three important variables in the cooperative ell as model. For each of the important parameters equilibrium: the carbon tax in 2000, the effi- d the of the model (see the description in Appendix cient reduction of CO₂ emissions in 2000, and e cen- A), we have varied the parameter by changing the change in global mean temperature in in the it from the subjective 50th percentile to the lhaus, subjective 90th percentile. 15 The exact deri- vation of the uncertainty range was developed the sensitivity analyses, we estimate the (subjective) 90th percentile of the distribution, Γ⁹⁰. Figure 10 shows the ratio of different outcomes for the 90th percentile of a ie sys- 15 Symbolically, we can represent the RICE model as a variables to the 50th percentile of that variable; that is, = Ras- mapping, Y, = F(X, Γ), where Y, is the vector of en- 4 = F(X,-,; r⁵⁰, where Д, is the ratio of 1975), dogenous and policy variables, X, is a vector of current outcomes for variables of interest when varying the ith vels of and lagged exogenous variables, and Γ is the set of un- parameter, Г.ˢ⁰ is the vector of Γ with all variables set at exem- certain parameters. The base run estimates outcomes for their 50th percentile while Γ,⁹⁰ is the vector of parameters ne de- the "best-guess" parameters (Γ⁵⁰, which represents the with all variables but the ith set at the 50th percentile while 50th percentile of the distribution of the parameters). In the ith parameter is set at its 90th percentile. 760 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 4.1 2.0 1.5 Ratio of variable to base value 1.0 X X 0.5 0.0 p Θ₁ 1 t1 (AA/A)₆ β t2 b₂ A. δₘ T₀ (AP/P)₂ T₀* M₀ δₖ b, σ₀ Parameter varied X Temperature (2100) Carbon tax (2000) Control rate (2000) FIGURE 10. SENSITIVITY TESTS FOR PARAMETERS Note: The variables on the horizontal axis are parameters of the RICE model as defined in Appendix A. The markers indicate the ratio of the outcome variable in the sensitivity case to the outcome for the base case. The sensitivity cases set the values of the variables at the subjective 90th percentile. The outcome variables are the optimal cooperative carbon tax in 2000, the optimal cooperative reduction rate for CO₂ emissions in 2000, and global mean temperature in 2100. 2100. For each of the three variables, we have variables are relatively unimportant for the displayed in Figure 10 the ratio of the value of results. 16 the variable in the sensitivity run to the value In analyzing model sensitivity, it is easy to of the variable in the base case. become lost in the details. For policy purposes, Figure 10 indicates that the results are ex- however, the single most critical question is tremely sensitive to the pure rate of social time how an uncertainty affects current policy, preference. The low rate of time preference which is best seen in the effect on the carbon (equal to 1 rather than 3 percent per year) in- tax. By this standard, the two crucial parame- creases the carbon tax by a factor of 4 and the ters are the discount rate (which indicates the control rate by a factor of almost 2. In addition, the damage intercept (which is the fraction of output lost from a doubling of atmospheric 16 These results parallel closely the findings of other CO₂) leads to a marked increase in both the studies on the sensitivity of policy to uncertainties about carbon tax and the control rate. The other major variables. MB VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 761 relative importance of the future compared to sions, and CO₂ concentrations as compared the present) and the damages from climate with the earlier DICE model is largely offset change (which measure the willingness to by revisions in estimated effects of other pay to prevent or slow climate change). It is greenhouse gases. As a result the estimated interesting to note that both major uncer- extent of global warming in the market case tainties involve human preferences rather by the year 2100-approximately 3°C- than pure questions of "fact" about the nat- differs little between the RICE model and ural sciences. other estimates. Second, the efficient or cooperative policies III. Conclusions in the regional model confirm estimates made in globally aggregated models, such as the To summarize, this paper has presented the DICE model. The best summary variable for RICE model, which is a new dynamic, multi- efficient controls is the carbon tax, which is region, general-equilibrium model of climate calculated to be about $6 per ton carbon in and the economy. It differs from earlier work, 2000, a number that is virtually identical to which focussed on a globally aggregated ap- estimates for the efficient policy in the DICE proach, by introducing production, consump- model. 18 The estimated degree of control in the tion, emissions, and damages for different RICE model is, however, estimated to grow regions. This approach compares three differ- somewhat more rapidly than in the DICE and ent strategies for the control of global warm- other models, with estimated efficient carbon ing: a market approach in which no climate taxes at the end of the next century near $27 change policies are taken, a global coopera- per ton carbon. tive approach in which all countries choose Third, the RICE model provides estimates climate-change policies to maximize global in- of the efficient control rates in different σ₀ comes, and a noncooperative or nationalistic regions as well. In the efficient solution, car- approach in which each country takes policies bon taxes are identical in all regions. The con- to maximize its own national income. These trol rates will differ, however, because of results are tentative and subject to revision. different costs of reducing CO₂ emissions. The 2000) Further work will be necessary to test their ro- estimates presented here indicate that the ef- bustness against alternative assumptions, to ficient emissions control rates will be highest appraise the results for different coalitions, and in China and the former Soviet Union and low- : markers to compare the results against other models. est in Japan and Europe, with the differences vity cases ive carbon Subject to these reservations, the following are being at least a factor of two. These results a 2100. the major conclusions. indicate that there will be substantial ineffi- First, the model produces results for the ciencies in any policy (such as that currently baseline (market or uncontrolled) which dif- in force under the Framework Convention) fer significantly from other projections. 17 that equalizes emissions control rates across Output and emissions in the RICE model are countries or does not allow trading of emis- for the estimated to grow much more rapidly than sions permits. in the DICE model or than in many inter- Fourth, a major contribution of this study easy to national projections (such as that of the In- is to estimate the difference between the ef- irposes, tergovernmental Panel on Climate Change). ficient policy and the noncooperative policy. stion is The more rapid growth comes largely from The noncooperative or nationalistic policy policy, a view of the growth process in which there is one in which countries maximize their carbon is considerable but incomplete convergence economic welfare taking policies of other arame- of per capita incomes of countries. The countries as given. This implies that small ates the higher projected growth of output, emis- countries, whose climate-change policies have little effect on their own economic of other 17 In the discussion that follows, the results for the :es about DICE model refer to DICE-123 as presented in Nordhaus 18 All dollar figures refer to prices in 1990 U.S. dollars (1994). at 1990 market exchange rates. 762 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VC welfare, will have little incentive to reduce Japan show reductions in cumulative dis- emissions while the largest countries will counted consumption until after the middle A, Pᵢ¹ have greatly attenuated incentives to engage of the next century. i in costly reductions in CO₂ emissions. The Seventh, the results indicate that there are O( calculations here indicate that the controls in major gains to taking an efficient cooperative the noncooperative case (as measured by the approach to coping with global warming as average rate of carbon tax) will be only 1/25 of opposed to the noncooperative approach. We the level of the cooperative case. That is, while estimate that the net economic gain from an α the average carbon tax in 2000 is estimated efficient policy has a discounted value of $344 b₁, to be about $6 per ton carbon in the coop- billion relative to the market scenario, while ß = € erative case, it is calculated to be about $0.24 the noncooperative policy has a gain of only γ per ton in the noncooperative case. More- $43 billion. Hence, there are clear gains to at- δₖ over, the divergence between the coopera- taining a cooperative policy (assuming, of δₘ tive and the noncooperative policies is course, that the policy is itself efficient). The I A = calculated to increase over time as the in- gains from cooperation would be even larger t equality of county sizes decreases, and this if climate change proved to have catastrophic P = divergence would increase further if large consequences that are very unevenly felt across ( countries like China, India, Russia, Canada nations. T1, f or the United States splinter into smaller In sum, the results of this new integrated F countries or decision-making units. model of climate and the economy empha- - Fifth, these results indicate that the stakes in sizes the implications of the fact that while 01, controlling global warming are modest in the climate change is a global externality, the context of overall economic activity over the decision makers are national and relatively next century. If our estimates are accurate, small. These inherent difficulties involved in they indicate that the losses from global warm- planning over a horizon of a century or more Cᵢ( about so uncertain and complex a phenom- cᵢ( ing will be in the range of 1 to 2 percent of CA global income over the next century. The net enon are compounded by the dispersed na- Dᵢ( costs (that is, climate-change damages less ture of the decisions and the strong tendency E,( mitigation costs) can be reduced by perhaps for free-riding by nonparticipants in any EX 1/3 percent of income by a judicious choice global agreement. Countries may therefore F(1 c of climate-change policies-although, to be be triply persuaded not to undertake costly Ω,( sure, the impact is much greater on our de- efforts today-first because the benefits are a scendants than on ourselves. According to so conjectural, secondly because they occur Kᵢ( RICE, successful cooperation would lead to so far in the future, and third because no in- IM, dividual country can have a significant im- M( net gains, but the failure to cooperate is un- ii likely to lead to economic disaster over the pact upon the pace of global warming. The NF. next century. present study indicates that the third of these, Φᵢ Sixth, the pattern of gains and losses the dispersed nature of the decision making Qi( from different strategies is quite surprising. and the consequent diluted incentives to act, R(1 T(t All countries gain from the noncooperative is a powerful hindrance to setting efficient 1. approach, although the amount of gain is climate-change policies. T*( relatively small. The net gains from coop- lo eration without international transfers are APPENDIX A: EQUATIONS OF THE RICE MODEL uᵢ(₁ W quite unevenly distributed, with the major This appendix gives the details of the RICE model. We S gains accruing to developing countries with first list and define the variables and then provide the com- Y,(; low and rapidly growing emissions. High- plete equation listing. d income countries have but modest gains to cooperation, but the United States actually 1. Variables loses from cooperating relative to a nonco- The variables are as follows. In the listing, 1 always operative strategy. In addition, the time path 1 refers to time (1 = 1990, 2000, ...) while i refers to the of gains and losses indicates that even in 1,(1 region (i = 1, , n = USA, Japan, Europe, ...). The the cooperative scenario, all regions except regional definition is given in Appendix B. VOL. 86 NO. 4 NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 763 Exogenous Variables. 2. Equations A,(t) = level of technology P₁(t) = population at time 1, also proportional to labor (A1) inputs O(t) = forcings of exogenous greenhouse gases = E E Parameters. subject to a = elasticity of marginal utility of consumption b1.1, b₂ = parameters of emissions-reduction cost function (A2) ß = marginal atmospheric retention ratio of CO₂ emissions (A3) γ = elasticity of output with respect to capital δₖ = rate of depreciation of the capital stock (A4) = Sm = rate of transfer of CO₂ from atmosphere to other reservoirs À = feedback parameter in climate model (inverse to temperature-sensitivity coefficient) P = pure rate of social time preference σ₁(t) = CO₂ emissions/output ratio T1, T2, T3, T4 = parameters of climate equation (T₁ is a (A5) function of the heat capacity of the atmosphere and up- per ocean while T2 depends upon the turnover time be- (A6) tween the upper ocean and the deep ocean) Θ₁,, Θ₂ = parameters of climate damage function (A7) S Endogenous Variables. C,(t) = total consumption (A8) c₁(t) = per capita consumption CA, (t) = current account balance (A9) T(t) = T(t - 1) D,(t) = damage from greenhouse warming E,(t) = CO₂ emissions + EXᵢ,(t) = exports from region i to region j F(t) = radiative forcing from all greenhouse gas concentrations Ωᵢ(t) = output scaling factor due to emissions controls (A10) and to damages from climate change K,(t) = capital stock IMᵢ.ⱼ(t) = imports from region i to region j (All) = 4.1 log[ M(t)/M(0)} log(2) + 0(t) M(t) = increase in mass of CO₂ in atmosphere from pre- industrial level NFA, (t) = net foreign assets of country i (A12) 1 i=1,2,...,n. Φ, = welfare weight on country i Q,(t) = gross domestic or regional product R(t) = net rate of return on capital Σ γQ,(t) T(t) = atmospheric temperature relative to preindustrial (A13) level T*(t) = deep ocean temperature relative to preindustrial level (A14) NFA₁(t) =NFA(t -1)+ CA(r- 1) uᵢ(t) = u,[c,(t)] = utility of per capita consumption W = social welfare function determined by country con- (A15) sumption levels " Y, (t) = gross national or regional product (net of climate damage and mitigation costs) (A16) -CA;(t) ≤ 0.1Q,(t) (A17) -NFA,(1) ≤ 0.1Q,(t) â Policy Variables. I,(t) = gross investment (A18) â (t) = rate of emissions reduction 764 THE AMERICAN ECONOMIC REVIEW SEPTEMBER 1996 VI G: APPENDIX B: REGIONAL GROUPING IN THE RICE MODEL Number of Gross domestic product Population, 1990 CO₂ emissions, 1990 Country or group countries (millions of 1990 US $) (thousands) (millions of tons C) In 1) United States 1 5,464,796 250,372 1,370.0 2) Japan 1 2,932,055 123,537 291.5 3) Former Soviet Union 1 855,207 289,324 1,065.7 4) China 1 370,024 1,133,683 805.5 5) Europe 1 6,828,042 366,497 872.3 6) Huge 1 295,760 849,515 215.4 In 7) Large 2 586,072 327,274 593.0 8) Midsized 11 2,155,910 442,370 789.7 9) Small 38 1,272,414 876,027 1,212.0 10) Tiny 137 318,464 607,503 623.9 M Total 21,078,746 5,266,102 7,839.1 Bottom 5 groups (ROW) 4,628,621 3,102,689 3,434.0 Selected countries in groups 6 through 10: M Gross domestic product Population, 1990 CO₂ emissions, 1990 Code Country (millions of 1990 US $) (thousands) (millions of tons C) SI India 295,760 849,515 215.4 S2 Brazil 479,214 149,042 317.1 Indonesia 106,859 178,232 275.9 S3 Canada 566,694 26,522 127.6 Australia 296,053 17,045 72.1 Mexico 244,046 81,724 144.1 Argentina 141,353 32,322 33.1 M Turkey 108,447 56,098 35.3 South Africa 101,963 37,959 78.0 S4 Venezuela 48,599 19,325 43.0 Romania 37,625 23,200 59.4 M Nigeria 35,460 96,203 95.9 Egypt 35,400 52,426 22.3 Slovenia 17,331 2,000 4.8 S5 Kenya 8,675 24,160 5.0 Iceland 6,024 255 0.5 Honduras 2,944 5,105 12.0 N: Maldives 174 214 0.0" Anguilla 23 7 0.0" Tuvalu 5 9 0.0" a Less than 50,000 tons per year. REFERENCES nization for Economic Cooperation and De- velopment, Paris, processed 1992. Bodansky, Daniel M. "Emerging Climate Change EMF, Energy Modeling Forum 14. "Background Regime." Annual Review of Energy and the Papers" and "Tabulations of Model Runs." Environment, 1995, 20, forthcoming. Unpublished, 1995. Brooke, Anthony; Kendrick, David and Meeraus, Fankhauser, Samuel. "The Economic Costs Alexander. GAMS: A user's guide. Red- of Global Warming: Some Monetary Esti- wood City, CA: The Scientific Press, mates," in Y. Kaya, N. Nakicenovic, W. D. 1988. Nordhaus, and F.L. Toth, Costs, impacts, and Dean, Andrew and Hoeller, Peter. Costs of re- benefits of CO₂ mitigation. Laxenburg, Aus- ducing CO₂ emissions: Evidence from six tria: International Institute for Systems Anal- global models, OCDE/GD(92) 140, Orga- ysis, Laxenburg, Austria, CP-93-2, 1993. NORDHAUS AND YANG: CLIMATE-CHANGE STRATEGIES 765 ΓEMBE and Weyant, John P. "EMF-12: Negishi, T. "Welfare Economics and the Ex- Comparisons of the Costs of Re- istence of an Equilibrium for a Competitive Emassions." American Economic Economy." Metroeconomica, 1960, 12, pp. May 1993, 83(2), pp. 318-23. 92-97. CO2 emissio (millions of Panel on Climate Change Nordhaus, William D. "An Intertemporal Climate change: The IPCC scien- General-Equilibrium Model of Economic 1,370,0 2913 Expersement. J. T. Houghton, G. J. Growth and Climate Change," in Wood, 1,065.7 and J.J. Ephraums, eds. New York: David, ed., Proceedings of a Conference at 8053 University Press, 1990. MIT, October 1990. 872.3 Energy Agency (IEA). Climate . "To Slow or Not to Slow: The Eco- 215.4 593,0 policy initiatives: 1994 update, Vol. nomics of the Greenhouse Effect." Eco- Organization of Economic Coop- nomic Journal, July 1991, 101(407), pp. 789.7 1,212.0 and Development, 1994. 920-37. 623.9 S. and Richels, Richard. Buying . "An Optimal Transition Path for 7,839.1 phouse insurance: The economic costs Slowing Climate Change." Science, No- 3,434.0 CO2 emission limits. Cambridge, MA: vember 20, 1992, pp. 1315-19. Press, 1992. . Managing the global commons: The S. and Rutherford, Thomas F. "In- economics of the greenhouse effect. Cam- CO2 emissions Gonal Trade. Capital Flows and Sec- bridge, MA: MIT Press, 1994. (millions of to Analysis: Formulation and Solution Nuclear Regulatory Commission. Reactor safety 215.4 Intertemporal Equilibrium Models," in study: An assessment of accident risks in 317.1 W. Cooper and A. B. Whinston, eds., U.S. Commercial nuclear power plants, 275.9 directions in computational econom- NUREG-75/014 WASH-1400. Washing- 127.6 Dordrecht. Netherlands: Kluwer, 1994, ton, DC: 1975. 72.1 191-205. Peck, Stephen C. and Teisberg, Thomas J. 144.1 33.1 Andreu: Whinston, Michael D. and "CETA: A Model for Carbon Emissions 5.3 Green, Jerry R. Microeconomic theory. New Trajectory Assessment." Energy Journal, 3.0 York: Oxford University Press, 1995. 1992, 13(1), pp. 55-77. 43.0 Morgan, M. Granger and Henrion, Max. Uncer- Wigley, T.M.L. "The Contribution from Emis- 59.4 tainty: A guide to dealing with uncertainty sions of Different Gases to the Enhanced 95.9 22.3 in quantitative risk and policy analysis. Greenhouse Effect," in Ted Hanisch, ed., 4.8 New York: Cambridge University Press, Climate change and the agenda for re- 5.0 1990. search. Boulder, CO: Westview Press, 0.5 12.0 National Academy of Sciences. Stratospheric 1994, pp. 193-222. 0.0' ozone depletion by halocarbons: Chemistry Yang, Zili. "Essays on the International As- 0.0' and transport. Report of a committee of the pects of Resource and Environmental 0.0" National Research Council, National Re- Economics." Ph.D. dissertation, Yale Uni- search Council, Washington, DC, 1979. versity, 1993. eration and Do- 1992. "Background Model Runs." onomic Costs Monetary Esti- enovic, W. D. 5, impacts, and xenburg, Aus- Systems Anal- 3-2, 1993.