Energy system decarbonization and productivity gains reduced the coupling of CO2 emissions and economic growth in 73 countries between 1970 and 2016

•We estimate impacts of five mechanisms on nations’ CO2 emissions over 1970–2016•Without these mechanisms, grow as fast the economy•Energy system decarbonization was primary mechanism for high-income countries•Productivity gains were low-income countries Fearing negative repercussions economic growth, climate policy is not track to meet Paris Agreement temperature target. However, academic literature disagrees both impact growth mitigation and effectiveness various reduction. Here, we investigate relationship between in 73 during period 1970–2016. We find that absence would have indeed grown at same rate economy. mechanisms—energy decarbonization, electrification, increased productivity, deindustrialization, winter warming—are identified successfully reducing by 19 gigatonnes, mostly periods growth. Yet, observations indicate reduction rates consistent with could be achieved while maintaining only if energy systems are more rapidly decarbonized. Nations must curtail carbon dioxide (CO2) 7% per annum targets. A perceived growth-climate trade-off has diminished political will act. there no scholarly consensus regarding magnitude a lack ex post evidence extent which measures can effectively lower emissions. present structural equation model integrating characteristics 1970–2016 empirically assess influence GDP-CO2 countries. Robust specifications statistical tests, found simple unitary scale effect capita GDP emissions, emission-reduction principally productivity gains, collectively contributed global emission reductions petagrams. Within observed year-to-year development, about 10% instances The world explosive human population unprecedented levels welfare past two centuries through improved utilization fossil fuels.1Fischer-Kowalski M. Schaffartzik A. Energy availability sources determinants societal development long-term perspective.MRS Sustain. 2015; 2: 1Google Scholar In 2015 Agreement, 192 agreed halt rise well below 2°, preferably 1.5°.2UNFCCCReport Conference Parties its Twenty-First Session [the 2015]. United Framework Convention Climate Change, 2016Google Achieving 1.5°C warming goal requires cutting greenhouse gas (GHG) than year average from 2020 on.3Höhne N. den Elzen Rogelj J. Metz B. Fransen T. Kuramochi et al.Emissions: World Has Four Times Work or One-Third Time.Nature. 579. Nature Publishing Group, 2020: 25-28https://doi.org/10.1038/d41586-020-00571-xGoogle latest UN synthesis nationally determined contributions (NDCs) under warns current commitments, even fully realized, far adequate target Agreement.4UNFCCCNationally Determined Contributions - Synthesis Report Secretariat. 2021Google Worse, been achieving their NDC pledges, GHG continued 2019 fell 6.4% owing COVID-19 pandemic.3Höhne Scholar,5Tollefson COVID curbed 2020—but much.Nature. 2021; 589: 343Google first step ratcheting up NDCs occur upcoming 26th conference parties (COP26) UNFCCC Glasgow November 2021. commitments negotiations invariably tied question fundamental how reconcile required attain most other sustainable goals tricky issues share cost remaining budget.6Höhne Den Escalante D. Regional targets based effort sharing: comparison studies.Clim. Policy. 2014; 14: 122-147Google Scholar,7Joy Tschakert P. Waisman H. Abdul Halim S. Antwi-Agyei Dasgupta al.Sustainable poverty eradication inequalities.in: Global Warming 1.5° C Special Report. Intergovernmental Panel 2018: 445-538Google Studies empirical development8Blanco G. Gerlagh R. S J B de Coninck H.C. Diaz Morejon C.F. Drivers, trends mitigation.in: Edenhofer O. Pichs-Madruga Sokona Y. Farahani E. Kadner Seyboth K. Change 2014: Mitigation Change. Contribution Working Group III Fifth Assessment Cambridge University Press, 351-411Google focused estimating CO2-GDP elasticity, i.e., percentage change country's every 1% increase (Figures 1A 1B ). They variously shown everything 0.3% decrease 2.5% percent GDP9Fernandez-Amador Francois J.F. Oberdabernig D.A. Tomberger Carbon growth: an assessment CrossMark production consumption inventories.Ecol. Econ. 2017; 135: 269-279https://doi.org/10.1016/j.ecolecon.2017.01.004Google Scholar, 10Schmalensee Stoker T.M. Judson R.A. emissions: 1950-2050.Rev. Stat. 1998; 80: 15-27https://doi.org/10.1162/003465398557294Google 11Heil M.T. Selden International trade intensity cross-country econometric analysis.J. Environ. Dev. 2001; : 35-49Google 12Cole M.A. Neumayer Examining demographic factors air pollution.Popul. 2004; 26: 5-21https://doi.org/10.1023/B:POEN.0000039950.85422.ebGoogle 13York Rosa E.A. Dietz STIRPAT, IPAT ImPACT: analytic tools unpacking driving forces environmental impacts.Ecol. 2003; 46: 351-365https://doi.org/10.1016/S0921-8009(03)00188-5Google 14Wagner Kuznets curve: cloudy picture emitted bad econometrics?.Resource 2008; 30: 388-408Google 15Managi Hibiki Tsurumi Does openness improve quality?.J. Manag. 2009; 58: 346-363Google 16Musolesi Mazzanti Zoboli panel data heterogeneous Bayesian estimation curves emissions.Appl. 2010; 42: 2275-2287Google 17Stern D.I. Between estimates emissions-income elasticity.Ecol. 69: 2173-2182https://doi.org/10.1016/j.ecolecon.2010.06.024Google 18Poumanyvong Kaneko urbanization lead less use emissions? analysis.Ecol. 70: 434-444Google 19Steinberger J.K. Roberts J.T. From constraint sufficiency: decoupling needs, 1975–2005.Ecol. 425-433Google 20Jorgenson A.K. Clark Are economy environment decoupling? comparative international study, 1960–2005.Am. Sociol. 2012; 118: 1-44Google 21Anjum Z. Burke P.J. Stern Modelling Emissions-Income Relationship Using Long-Run Growth Rates. Centre Economics Policy, Crawford School Public ANU, 2014Google 22Burke Shahiduzzaman short run: matter.Glob. Chang. 33: 109-121Google 23Liddle What elasticities income population? Bridging STIRPAT EKC via robust estimates.Glob. 31: 62-73Google 24Adewuyi A.O. Effects public private expenditures pollution: dynamic analysis.Renew. Rev. 2016; 65: 489-506Google 25Franzen Mader Predictors national do matter?.Climatic 139: 491-502Google 26Aklin Re-exploring nexus diffusion pollution.Environ. Resource 64: 663-682Google 27Fernández-Amador 269-279Google (for details, see Note S1 Table supplemental information). As result disparate findings, policymakers able rely analysis weighing trade-offs potentially conflicting social goals. Variation attributed sample-selection (see Figure 1B) potential spurious results given insufficient consideration statistically problematic properties investigated (i.e., non-stationarity,28Barros C.P. Gil-Alana L.A. De Gracia F.P. Stationarity long range dependence disaggregated data.Environ. 63: 45-56Google endogeneity,26Aklin Scholar,29Frankel J.A. Rose Is good environment? Sorting out causality.Rev. 2005; 87: 85-91https://doi.org/10.1162/0034653053327577Google cross-sectional heterogeneity14Wagner Scholar,30Vollebergh H.R.J. Melenberg Dijkgraaf Identifying reduced-form relations data: case pollution income.J. 27-42https://doi.org/10.1016/j.jeem.2008.12.005Google Scholar). hypothesize, meanwhile, failure studies identify monocausal lies separate independent causal historical pathways. recent success selected reduce noted prominently literature.33Olivier J.G.J. Janssens-Maenhout Muntean Peters J.A.H.W. Trends Emissions; 2016 Hague: PBL Netherlands Environmental Agency. Ispra: European Commission, Joint Research Centre, 2016http://edgar.jrc.ec.europa.eu/news_docs/jrc-2016-trends-in-global-co2-emissions-2016-report-103425.pdfGoogle 34JRC/ P.B.L. Emissions Fossil Fuel Use Cement Production 1970-2015 (EDGARv4.3.2). (JRC)/PBL Emission Database Atmospheric (EDGAR), 35Feenstra R.C. Inklaar Timmer M.P. next generation Penn Table.Am. 105: 3150-3182https://doi.org/10.1257/aer.20130954Google understanding enabled such hopeful development. Existing knowledge anthropogenic impeding predominantly derived using ante analyses.36Riahi Van Vuuren D.P. Kriegler Edmonds O’Neill B.C. Fujimori Bauer Calvin Dellink Fricko shared socioeconomic pathways energy, land use, implications: overview.Glob. 153-168Google 37Rogelj Shindell Jiang Fifita Forster Ginzburg V. Handa C. Kheshgi Kobayashi compatible context development.in: 1.5°C. (IPCC), 93-174Google 38IEANet Zero 2050: Roadmap Sector. Agency, Ex analyses predict modeling presumed relationships of, e.g., solar power deployment, transformations, Empirical shown, post, developments differ qualitatively quantitatively engineering estimates, confirming disproving hypothesized adopted studies.39Carleton T.A. Hsiang S.M. Social climate.Science. 353: aad9837Google Scholar,40York Do alternative displace fuels?.Nat. Clim. 441Google Moreover, highly developed posited curve (EKC) hypothesis.13York Scholar,41Holtzeakin Stoking fires—CO2 economic-growth.J. 1995; 57: 85-101https://doi.org/10.1016/0047-2727(94)01449-XGoogle 42López-Menéndez A.J. Pérez Moreno costs renewable energy: re-visiting curve.J. 145: 368-373Google 43Jebli M.B. Kahia interdependence non-renewable energies, service development: 65 countries.Climatic 2020; 162: 193-212Google It suggests people willingness pay quality once they become sufficiently affluent. Recent studies, however, disputed methodological robustness essentially undermined validity hypothesis explanation growth.14Wagner Scholar,44Stern after 25 years.J. Bioeconomics. 19: 7-28https://doi.org/10.1007/s10818-017-9243-1Google Another declining accompanying offshoring heavy industries trade, but remains contradictory.25Franzen Scholar,45Levinson Taylor M.S. Unmasking haven effect.Int. 49: 223-254Google propose role additional mediating apply it extensive dataset covering economies (detailed S2 S2, show that, keeping constant, lockstep GDP. test suggested our identifies following influential: (1) increases (2) (including increasing shares fuel switching coal gas), (3) (4) warming, (5) deindustrialization selection experimental procedures). Over 47 years, 28%) high when electrification saturated, low levels, 36%) addition being driver Therefore, study resurrects debate plays vital contribute reductions. Our findings also inform around showing without country likely achieve annually gradually 2% (Tanzania) 17% (Qatar) annual reach net-zero 2050. valuable informing policies, few years left before exceed budget.46Pachauri R.K. Allen M.R. Barros V.R. Broome Cramer W. Christ Church Clarke L. Dahe Q. Groups I, II Ipcc, benchmark model, associated average, holding all constant (Table 1; full available S3 S3, this elasticity specifications. account previous asymmetric effects expansions versus recessions, related countries' three-equation (CO2, GDP) simultaneous model. addition, models estimated 5-year rolling windows eliminate cyclical variations sources. Results tests S4 Tables S4–S6, information.Table 1The economic, system, temperature-related modelFactors affecting emissionsaFull provided including non-statistically significant (p ≥ 0.05), equation, SEs CIs bootstrapping.Explanatory variablesCoefficientsBootstrap SE95% Bca CIEconomic Economic growthΔbΔ difference operator. GDP/p1.0194∗∗∗cThe different 1 (prob > χ2 = 0.81).0.10820.81501.2406(1)dChanges desirable directions, indicated coefficients, correspond (1)–(5) referred throughout manuscript. Increases productivityΔ A∙L/p−0.4824∗∗∗0.0923−0.6764−0.3184(2) Deindustrialization- Δ Vsh_Ind−0.7724∗∗∗0.2890−0.3107−1.4585Energy transition(3) Decarbonization renewablesΔ TPESsh_renw−1.4942∗∗∗0.1333−1.7125−1.1761 Coal natural gas- TPES_FFsh_coal−0.5883∗∗∗0.1063−0.3568−0.7541(4) ElectrificationΔ Electrification−1.2589∗∗∗0.6091−2.4193−0.4669Other(5) Winter warmingΔ Tmin−0.0050∗∗∗0.0007−0.0065−0.0037n (countries/regions)73N (observations)3,283R20.45∗p < 0.05, ∗∗p 0.01, ∗∗∗p 0.001 (two-sided Z test, 1,000 bootstrap runs).a Full bootstrapping.b operator.c 0.81).d Changes Open table new tab ∗p runs). changes capita, describing economy, significantly affect sample 1). drives time overall roughly 0.5% Combining growth-inducing direct causes net 0.5%. Such growth-driving S3), indicating type nation's pathway. Consistent conventional wisdom deindustrialization—for example, move services—reduces reduced 0.8% non-industrial output grew 1%. further reveal majority (>80%) attributable S5). used supply (referred “renewables share” thereafter) 1.5% yearly reveals two-thirds shares' lowering intensity, rest improving efficiency Specifically, renewables “CO2 equation”) “energy equation”); (CO2 equation). efficient fuels serving demand observation period. accounting approaches calculating content non-fossil fuels.47IEAWorld Balances. 2019Google Among fuels, displacing switching) decreased 0.6% substituted revealed S5), much efficiency. coal, 0.4% equation), 0.2% (energy With mix, Every electricity final 1.3% showed coldest month 1°C Finally, 0.02%–0.04% 0.05) time-invariant emerging (Bulgaria, France, Ireland, Poland, Romania, Sweden, UK) Mozambique. be, instance, institutions48Laegreid O.M. Povitkina institutions moderate relationship?.Ecol. 2018; 441-450Google geography, culture, potentials.49Eskander Fankhauser Reduction legislation.Nat. 10: 750-756Google All included had changing sample. precisely 0.001) despite variation across levels. For magnitudes depended processes, illustrated relatively greater confidence intervals (CIs) 2A. goodness-of-fit measured R2, 0.45, compared R2 statistics 0.05 0.27 prior examining capita.20Jorgenson Scholar,22Burke Scholar,50Lohwasser Schaffer Brieden drivers traditional standardized 178: 106811Google Scholar,51Casey Galor faster growth.Environ. Res. Lett. 12: 014003Google Statistical (Note S4–S6 information) verified multicollinearity, non-stationarity, correlation S5 S7–S11, To quantify total main 1970–2016, combined year-by-year (3,283 country-year observations, Equation 7 le