Business Cycles & CO 2 Emissions Baran Doda Grantham Research Institute, LSE 26 March 2015
Context & motivation We know relatively little about the relationship between uctuations in GDP and CO 2 emissions. The nature of the relationship is important for making sure theoretical models resemble the real world carbon pricing policy instruments in practice
Context & motivation We know relatively little about the relationship between uctuations in GDP and CO 2 emissions. The nature of the relationship is important for making sure theoretical models resemble the real world carbon pricing policy instruments in practice
Research topic 1. What are the key properties of uctuations in GDP and CO 2 emissions in a given country? US GDP & Emissions Time Series & Trend Components Cyclical Components GDP 14 14.5 15 15.5 16 1950 1960 1970 1980 1990 2000 2010 GDP Emissions 13.4 13.6 13.8 14 14.2 EMIS emis.06.04.02 0.02.04.06.04.02 0.02.04 gdp 2. Is there a systematic relationship between these properties and GDP per capita? (Annual data from 122 countries covering 1950-2011)
Research topic 1. What are the key properties of uctuations in GDP and CO 2 emissions in a given country? US GDP & Emissions Time Series & Trend Components Cyclical Components GDP 14 14.5 15 15.5 16 1950 1960 1970 1980 1990 2000 2010 GDP Emissions 13.4 13.6 13.8 14 14.2 EMIS emis.06.04.02 0.02.04.06.04.02 0.02.04 gdp 2. Is there a systematic relationship between these properties and GDP per capita? (Annual data from 122 countries covering 1950-2011)
Business cycle properties of emissions FACT 1: Emissions are procyclical in a typical country. Table FACT 2: Procyclicality of emissions is greater in countries with higher GDP per capita. Table Figure FACT 3: Emissions are cyclically more volatile than GDP in a typical country. Table FACT 4: Cyclical volatility of emissions is greater in countries with lower GDP per capita. Table Figure a Figure b
Business cycle properties of emissions FACT 1: Emissions are procyclical in a typical country. Table FACT 2: Procyclicality of emissions is greater in countries with higher GDP per capita. Table Figure FACT 3: Emissions are cyclically more volatile than GDP in a typical country. Table FACT 4: Cyclical volatility of emissions is greater in countries with lower GDP per capita. Table Figure a Figure b
Business cycle properties of emissions FACT 1: Emissions are procyclical in a typical country. Table FACT 2: Procyclicality of emissions is greater in countries with higher GDP per capita. Table Figure FACT 3: Emissions are cyclically more volatile than GDP in a typical country. Table FACT 4: Cyclical volatility of emissions is greater in countries with lower GDP per capita. Table Figure a Figure b
Business cycle properties of emissions FACT 1: Emissions are procyclical in a typical country. Table FACT 2: Procyclicality of emissions is greater in countries with higher GDP per capita. Table Figure FACT 3: Emissions are cyclically more volatile than GDP in a typical country. Table FACT 4: Cyclical volatility of emissions is greater in countries with lower GDP per capita. Table Figure a Figure b
Source: World Bank (2014)
What are the facts good for? Implications for xed instruments: Carbon tax = emissions uncertainty Emissions trading system = cost uncertainty Improving instrument performance Indexed regulation
What are the facts good for? Implications for xed instruments: Carbon tax = emissions uncertainty Emissions trading system = cost uncertainty Improving instrument performance Indexed regulation
How to price carbon in good times...and bad! Two broad policy implications: 1) Making the stringency of regulation responsive to economic uctuations can decrease overall burden of regulation. 2) Whatever instrument is chosen to price carbon, it should apply to as large a group of emitters as possible. Linking previously independent ETSs can mimic responsive regulation and lead to overall gains. However, country-specic gains are not assured.
How to price carbon in good times...and bad! Two broad policy implications: 1) Making the stringency of regulation responsive to economic uctuations can decrease overall burden of regulation. 2) Whatever instrument is chosen to price carbon, it should apply to as large a group of emitters as possible. Linking previously independent ETSs can mimic responsive regulation and lead to overall gains. However, country-specic gains are not assured.
How to price carbon in good times...and bad! Two broad policy implications: 1) Making the stringency of regulation responsive to economic uctuations can decrease overall burden of regulation. 2) Whatever instrument is chosen to price carbon, it should apply to as large a group of emitters as possible. Linking previously independent ETSs can mimic responsive regulation and lead to overall gains. However, country-specic gains are not assured.
Conclusions Business cycle properties of emissions dier across countries. Climate change policy design and performance can be improved by conditioning policy on business cycle uctuations cross-country dierences in business cycle properties of emissions other country characteristics
Business Cycles & CO 2 Emissions Baran Doda Grantham Research Institute, LSE 26 March 2015 Email:
Table: Cyclicality of emissions Mean Std. Dev. Min Max Sample ρ ey 0.297 0.244 0.305 0.824 Full (N=122) ρ ey 0.260 0.229 0.305 0.725 Restricted (N=89) Notes: A bar over a variable indicates a sample mean. The null hypothesis that ρ ey is equal to zero tested against a two-sided alternative in each case, where * implies p<0.10, ** implies p<0.05, and *** implies p<0.01.
Table: Cyclicality of emissions across countries Value Sample ρ(ρ ey,gdppc 2009 ) 0.327 Full (N=122) ρ(ρ ey,gdppc 2009 ) 0.359 Restricted (N=89) Notes: The null hypothesis that ρ(ρ ey,gdppc 2009 ) is equal to zero tested against a two-sided alternative in each case, where * implies p<0.10, ** implies p<0.05, and *** implies p<0.01.
Table: Volatility of emissions Mean Std. Dev. Min Max Sample σ e 0.078 0.064 0.018 0.358 σ y 0.029 0.018 0.006 0.109 Full (N=122) σ rel 3.040 2.492 0.701 17.221 σ e 0.068 0.051 0.018 0.285 σ y 0.023 0.011 0.009 0.081 Restricted (N=89) σ rel 3.082 2.197 1.019 15.258 Notes: A bar over a variable indicates a sample mean. In the last row of each panel the null hypothesis that σ rel = 1 is tested against the alternative that σ rel > 1, where * implies p<0.10, ** implies p<0.05, and *** implies p<0.01.
Table: Volatility of emissions across countries Value Sample ρ(σ e,gdppc 2009 ) 0.220 Full (N=122) ρ(σ e,gdppc 2009 ) 0.316 Restricted (N=89) ρ(σ rel,gdppc 2009 ) 0.203 Full (N=122) ρ(σ rel,gdppc 2009 ) 0.235 Restricted (N=89) Notes: The null hypothesis that a given correlation coecient is equal to zero is tested against a two-sided alternative in each case, where * implies p<0.10, ** implies p<0.05, and *** implies p<0.01.
Figure: Procyclicality of emissions increases with GDP per capita (Fact 2) Cyclicality (ρey).5 0.5 1 COD NER MDG MWI ZMB ZWE TZA ETH MLI KEN CIV BFA UGA BGD CMR SEN IRQ TJK AGO NGA GHA PHL PAK MOZ KHM BOL IND SDN VNM MDA GTM JOR DZA TKM BHR ZAF LKA MAR ROU CHN HRV BLR AZE ALB BRA LVA CYP KWT UKR TUR ARM CZE PER ARG CHL GEO SVK MKDIDN JAM YEMMMR EGY LCA UZB DOM ECU IRN TUN COL MEX POL THA URY KAZ MYS HUN BGR CRI SYR OMN SAU BRB VENQAT 6 8 10 Log GDP per capita in 2009 (GDPpc2009) Note: 3 letter country codes in red denote members of the full but not the restricted sample. Regression line drawn for the full sample. KGZ BIH RUS LTU MLT PRT ARE GRC SVN ESP EST ITA ISR NZL JPN BEL GBR AUT DEU FRA TWN CHE AUS KOR ISL CAN NLD FINIRL SGP DNK SWE TTO NOR USA HKG LUX
Figure: Volatility of emissions decreases with GDP per capita (Fact 4) Volatility (σe) 0.1.2.3.4 COD NER CMR AGO BFA NGA TJK ETH MDG IRQ SEN CIV TZA MLI KEN GHA ZMB ZWE UGA MWI BGD ARE QAT KHM IRN OMN SGP GEO KWT TTO SAU BHR BIH LCA ALB KGZ SYR YEM JAM SDN ECU ARM DZAMDA CHN VNM BRB MLT MOZ MMR LKA DOM TKM URY BOL CRI FIN IDN VEN CYP EST MAR EGY GTM PER JOR THA BGR MYS LTU ISR DNK ISL AZE PHL UKROU HKG TUN LVACHL SWE IRL CHE NOR MKD UZB KOR TWN COL MEX TUR HUN ARGKAZ PRT GRC ESP AUT NZL NLD BRA SVK JPNBEL FRA ZAF HRV CZE POL BLR SVN ITA PAK DEU CAN IND RUS GBRAUS USA LUX 6 8 10 Log GDP per capita in 2009 (GDPpc2009) Note: 3 letter country codes in red denote members of the full but not the restricted sample. Regression line drawn for the full sample.
Figure: Relative volatility of emissions decreases with GDP per capita (Fact 4) Relative volatility (σrel) 0 5 10 15 20 COD NER CMR TZA BFA MDG ETH SEN CIV BGD TJK NGA KEN GHA ZMB MLI AGO UGA ZWE MWI IRQ YEM SGP ARE TTO BHR LCA KHM SAU QAT LKA IRN ECU JAM BRB BOL GTM MLT NOR DNK VNM EGY FRA MMR SWE ALBGEO FIN CRI MAR OMN BEL MOZ DOMUZB COL AUT NLD IRL CHE PHL CHN KGZ SYRTHAMYS URY DZA IDN SDN ZAF BGR ARM PRT ESP PAK TUN HUN GRC ITA ISR JPN VEN KWT TWN ISL AUS MDA MKD TKM ROU HKG PER JOR BRA MEX NZL DEUGBR CAN TUR KAZ SVK CZE BIH POL CHL KOR IND ARGCYP SVN EST USA UKR HRV LTU RUS AZE LVA BLR LUX 6 8 10 Log GDP per capita in 2009 (GDPpc2009) Note: 3 letter country codes in red denote members of the full but not the restricted sample. Regression line drawn for the full sample.