Carbon bias of tariffs: are fossil fuels the culprits?

Cecilia Bellora

CEPII

Lionel Fontagné

PSE

Christophe Gouel

INRAE

CEPII

Youssef Salib

PSE, Ecole Nationale des Ponts et Chaussées

CEPII

Introduction

Trade and climate transition

  • Climate urgency \(\Rightarrow\) Reconsider the consistency of policies wrt climate objectives
  • In absence of uniform carbon pricing, trade policy structure affects emissions
  • Are current trade policies consistent with climate objectives?

Shapiro (2021, QJE)

The Environmental Bias of Trade Policy

  • Econometrics and simulation results
  • Econometrics results for manufacturing only:
    • Bias of trade policy toward carbon intensive goods
    • Implicit carbon subsidy of 40$/tCO\(_2\)
  • Political economy explanation:
    • Upstream sectors are more polluting
    • Upstream sectors have less tariffs because of lobbying

Shapiro (2021, QJE)

  • Quantitative model with all sectors:
    • Removing sectoral heterogeneity in bilateral relationships
    • 3.6% reduction in world emissions with tariffs and non-tariff barriers
    • 1.75% reduction with tariffs only (more than EU-ETS)
    • No reduction in welfare
    • Grounds for reform?

Other contributions on the carbon bias of trade policies

  • All sectors following Shapiro (2021)
    • Econometrics
      • Chepeliev and Corong (2022), extension to all GHG \(\Rightarrow\) smaller bias
      • Moreira and Dolabella (2024), all GHG and sectors in Latin America: highly heterogeneous bias
    • Modeling
      • Klotz and Sharma (2023), similar model but different accounting of emissions and data \(\Rightarrow\) 10 times smaller bias
  • Agriculture: Laborde et al. (2020) and Guerrero et al. (2022) show that agricultural tariffs decrease emissions

Our contribution

  • What we do
    • Quantitative trade model (Shapiro’s model + extensions)
    • Counterfactual trade policy simulation (tariffs only)
    • Focus on emission implications
  • Dissecting Shapiro’s result
    • Extension 2007–19 and all GHG
    • More recent data
    • Sectoral analysis
    • Fossil fuel extension:
      • Finite natural resources
      • Domestic taxes
  • Results
    • The bias is present for all years
    • It is much smaller when all GHG are included
    • The low tariffs on fossil fuels explain nearly all the bias
    • With the extensions, the bias is abolished or reversed

Data

4 main sources of data

  • EXIOBASE 3 world input-output tables 2007–19
    • Also used for GHG emissions other than CO\(_2\) from fossils
  • MAcMap-HS6 tariffs database
    • Every 3 years: 2007, 2010, 2013, 2016, 2019
    • Interpolated for other years
    • MFN and preferential tariffs in ad valorem equivalent
  • IEA world energy balances for energy data (yearly)
  • GTAP 11 for oil/gas/coal natural resource rents

Descriptive statistics

Sector Average tariff (%) % of imports in sector aggregate supply CO2 intensity (kg/€) Non-CO2 GHG intensity (kg/€)
Agriculture 6.82 13.46 0.13 1.19
Fossil Extraction 0.31 47.09 0.69 1.28
Brown Industries 1.93 25.11 0.31 0.12
Manufacturing n.e.s. 2.98 28.56 0.07 0.00
Services 0.00 3.78 0.17 0.01

Model

Model general description

  • NQTM, based on Caliendo and Parro (2015)
    • Parsimonious CGE, calibrated in exact hat algebra
    • Armington type of model
  • Production
    • With labor (value added) and intermediate goods
    • Cobb–Douglas, constant returns to scale
  • Initial equilibrium calibrated on EXIOBASE
  • Trade elasticities from Shapiro (2021)
  • 2 aggregations
    • 10 regions, 21 sectors (Shapiro’s aggregation)
    • 23 regions, 47 sectors

Households

  • Cobb–Douglas preferences over composite final goods: \[U_{j}=\prod_{k\in\mathcal{K}}\left(D_{j,k}^{\text{FC}}\right)^{\theta_{j,k}^{U}}\]
    • \(D_{j,k}^{\text{FC}}\): final demand for composite good \(k\) in country \(j\)
    • \(\theta_{j,k}^{U}\): share of expenditure spent on \(k\)
  • CES Composite final goods: \[D^{\text{FC}}_{j,k}=\left[\sum_{i\in\mathcal{I}}\beta_{ij,k}^{1/\sigma_{k}}\left(D_{ij,k}^{\text{FC}}\right)^{{(\sigma_{k}-1)}/{\sigma_{k}}} \right]^{{\sigma_{k}}/{(\sigma_{k}-1)}}\]
    • \(\beta_{ij,k}\): demand shifter
    • \(\sigma_{k}>1\): elasticity of substitution between varieties

Production costs

  • Production combines labor and a composite intermediate good according to a Cobb–Douglas
  • Unit cost : \[c_{i,l}= \left(\frac{w_{i}}{\theta_{i,l}^w} \right)^{\theta_{i,l}^w}\prod_{k\in \mathcal{K}} \left(\frac{p_{i,k}}{\theta_{i,kl}}\right)^{\theta_{i,kl}}\]
    • \(w_{i}\): wage
    • \(\theta_{i,kl}\in[0,1]\) input-output coefficients
    • \(\theta^w_{i,l}=1-\sum_{k\in\mathcal{K}}\theta_{i,kl}\) cost share of labor

Trade

  • Trade costs

    • Iceberg trade costs: \(\tau_{ij,k}\)
    • Tariffs: \(T_{ij,k}=1+t_{ij,k}\)

    \[p_{ij,k}=T_{ij,k} \tau_{ij,k} c_{i,k}\]

  • Same Armington with elasticity \(\sigma_k\) for final and intermediate demands

  • Gravity equation

    \[T_{ij,k}X_{ij,k}=\beta_{ij,k}\left(p_{ij,k}/p_{j,k}\right)^{1-\sigma_k}E_{j,k}\]

Model equations (in EHA, \(\hat{x}=x'/x\))

\[\begin{align} \label{eq:40} \hat{p}_{ij,k}&:\hat{p}_{ij,k}=\hat{T}_{ij,k}\hat{\tau}_{ij,k}\hat{c}_{i,k},\\ \label{eq:53} \hat{c}_{i,l}&: \hat{c}_{i,l}= \prod_{f\in \mathcal{F}} \left(\hat{w}_{i,f}\right)^{\theta_{i,f,l}^w}\prod_{k\in \mathcal{K}} \left(\hat{p}_{i,k}\right)^{\theta_{i,kl}},\\ \label{eq:63} \hat{w}_{j,f}&:w_{j,f}L_{j,f}\hat{w}_{j,f}=\sum_{l\in\mathcal{K}}\theta_{j,f,l}^w R_{j,l}\hat{c}_{j,l}\hat{Q}_{j,l},\\ \label{eq:62} \hat{E}_{j,k}&:E_{j,k}\hat{E}_{j,k}=\theta_{j,k}^{U}GNE_{j}\widehat{GNE}_{j} +\sum_{l\in\mathcal{K}}\theta_{j,kl}R_{j,l}\hat{c}_{j,l}\hat{Q}_{j,l},\\ \label{eq:48} \hat{Q}_{i,k}&:\hat{c}_{i,k}\hat{Q}_{i,k}=\sum_{j\in\mathcal{I}}\theta^{R}_{ij,k}\hat{X}_{ij,k},\\ \label{eq:61} \hat{X}_{ij,k}&:\hat{T}_{ij,k}\hat{X}_{ij,k}=\hat{p}_{ij,k}^{1-\sigma_{k}}\hat{p}_{j,k}^{\sigma_{k}-1}\hat{E}_{j,k},\\ \label{eq:p} \hat{p}_{j,k}&:\hat{p}_{j,k}=\left(\sum_{i\in\mathcal{I}}\theta_{ij,k}^{X}\hat{p}_{ij,k}^{1-\sigma_{k}}\right)^{{1}/{(1-\sigma_{k})}},\\ \label{eq:75} \widehat{GNE}_{j}&:GNE_{j}\widehat{GNE}_{j}=GDP_{j}\widehat{GDP}_{j}+\Delta_{j}\hat{\Delta}_{j}\\ \label{eq:6} \widehat{GDP}_{j}&:GDP_{j}\widehat{GDP}_{j}=\sum_{f\in\mathcal{F}}w_{j,f}L_{j,f}\hat{w}_{j,f} +\sum_{i\in\mathcal{I}}\sum_{k\in\mathcal{K}}t'_{ij,k}X_{ij,k}\hat{X}_{ij,k} \end{align}\]

GHG accounting

  • CO\(_2\) emissions from fossil fuel combustion
    • Associated to extracted fossil fuels
    • Emissions are proportional to extracted volume of coal, oil, and gas
    • Similar to Shapiro (2021)
    • Recalculated emission factor for each location (IEA data)
  • Other GHG emissions proportional to change in:
    • Production in volume for production sectors
    • Consumption in volume for final consumption

The exogenous shock

  • Harmonize tariffs between sectors: \(\bar{t}_{ij} = \sum_{k\in\mathcal{G}} t_{ij,k}X_{ij,k}/\sum_{k\in\mathcal{G}} X_{ij,k}\)
  • One tariff level (the pre-shock average) for all imports from country \(i\) to \(j\).
Tariff shock by main sector in reference scenario
Sector Average tariff shock (pp)
Agriculture -4.44
Fossil Extraction 2.21
Brown Industries 0.69
Manufacturing n.e.s. -0.44

Results

Shapiro replication and extension

Shock emission % change when considering only CO2 or all GHG
Scenario CO2 All GHG CO2 part Other GHG part
1. Shapiro (2021): Shapiro’s data, shock, and aggregation, year = 2007 -1.75 -1.03 -1.25 0.22
2. Updated economic calibration data (excluding CO2) -1.77 -1.18 -1.26 0.08
3. Update calibration data (including CO2) -1.50 -0.98 -1.06 0.08
4. No shock on services -1.72 -1.16 -1.22 0.05
5. Our sectoral aggregation (10 regions, 47 sectors) -1.38 -0.52 -0.98 0.46
6. Our sectoral and geographical aggregation (23 regions, 47 sectors) -1.68 0.31 -1.19 1.49
7. Our benchmark results: our data, shock, and aggregation, year = 2019 -2.02 -0.58 -1.40 0.82

Sector contributions

  • Change of tariff for only one sector. Same shock: \(\bar{t}_{ij} = \sum_{k\in\mathcal{G}} t_{ij,k}X_{ij,k}/\sum_{k\in\mathcal{G}} X_{ij,k}\)
  • All other sectors keep their tariffs
Sectoral shock effect on GHG emission and underlying the contribution of CO2 and Other GHG
Sector All GHG CO2 Other GHG
Agriculture (A) 0.61 -0.17 0.78
Fossil Extraction (F) -1.27 -1.18 -0.09
Brown Industries (B) -0.17 -0.15 -0.02
Manufacturing n.e.s. (M) 0.27 0.10 0.17
All -0.58 -1.40 0.82
All except Fossil Extraction 0.70 -0.20 0.91
A + F + B + M -0.56 -1.40 0.85
A + B + M 0.72 -0.22 0.94
Sector All GHG CO2 Other GHG
Fossil Extraction (F) -1.27 -1.18 -0.09
Coal (C) -0.39 -0.36 -0.03
Oil (O) -0.78 -0.73 -0.06
Gas (G) -0.09 -0.09 -0.01
C + O + G -1.27 -1.18 -0.09

Results by year

Upstreamness?

Extensions on fossil impact

Extension 1: Finite natural
resources

  • In our model, fossil extraction only needs labor and intermediate goods
    • Natural resources and rents are ignored
    • Tariffs increase local production in an unrealistic way
  • Extension to include natural resources in the production function
    • Value added is split between natural resource rents and other value added
    • Production function remains Cobb–Douglas
    • Rent = 0 for all sector except fossil extraction

Extension 1: Results

Sector Standard model Finite NR
Agriculture (A) 0.61 0.41
Manufacturing n.e.s. (M) 0.27 0.22
Fossil Extraction (F) -1.27 -0.53
Brown Industries (B) -0.17 -0.11
Coal (C) -0.39 -0.19
Oil (O) -0.78 -0.30
Gas (G) -0.09 -0.04
All -0.58 -0.04
All except Fossil Extraction 0.70 0.49

Extension 2: Domestic taxes as tariffs

  • For a non-producer:
    • Tax on fossil = uniform tariff (on all trade partners)
    • Taxes can have a trade reason: rent-extraction
  • Special taxes on fossil (energy, CO\(_2\)) are often very high (20–50% on average for non-producers).
  • Unique sector: concentrated production + special taxation
  • What happens if we include these taxes as tariffs?
Ad valorem equivalent net tax (%) on fossil fuels for the regions with import shares above 99%
Region Coal Oil Gas
Belgium, Netherlands, and Luxembourg 82.27 23.84 -
Eastern EU countries in OECD data - 50.05 -
France 31.77 78.47 29.93
Germany - 56.60 -
Italy 69.71 - -
Japan 11.99 23.60 83.89
Korea 66.48 24.90 36.55
Non EU European countries - 49.83 17.39
Nordic countries in the EU 59.23 - -
Norway 7.37 - -
Other EU countries in OECD data - 56.53 65.41
Spain - 55.83 49.14

Note: “-” indicates that for this fuel the region’s import share is below 99%.

Sources: Taxation calculated using OECD and EXIOBASE and net imports calculated from BP.

Extension 2: Results

% change in GHG emissions
Sector Fossil tax not included (benchmark results) With fossil tax for net imports > 99%
All sectors -0.58 2.14
Agriculture 0.61 0.65
Fossils -1.27 1.25
Brown Industries -0.17 -0.12
Manufacturing n.e.s. 0.27 0.36

Conclusion

  • The carbon bias uncovered by Shapiro (2021) confirmed for 2007–19
  • The bias is mostly due to the low of tariffs on fossils fuels
  • Bias
    • abolished by accounting for natural resources in fossil fuel production
    • and reversed by accounting for domestic taxes in non-producing countries
  • Bias very sensitive to simple modeling hypotheses
    • No robust ground for a large tariff reform

Thank You !

Any question ?

Appendix

Model with nonlinearities

% change in GHG emissions
Model Parameterization (σU, σQ, σIC, σVA) All Fossils
Standard model Benchmark (1, 1, 1) -0.58 -1.27
as Baqaee-Fahri, except for σVA (0.9, 0.5, 0.2) -0.60 -1.21
low elasticities (0.9, 0.1, 0.2) -0.61 -1.17
very low elasticities I (0.9, 0.05, 0.05) -0.61 -1.16
very low elasticities II (0.1, 0.05, 0.05) -0.59 -1.09
Finite natural resources Benchmark (1, 1, 1, 1) -0.04 -0.53
as Baqaee-Fahri (0.9, 0.5, 0.2, 1) 0.00 -0.46
as Baqaee-Fahri, except for σVA (0.9, 0.5, 0.2, 0.5) 0.09 -0.36
low elasticities (0.9, 0.1, 0.2, 0.5) 0.13 -0.30
very low elasticities I (0.9, 0.05, 0.05, 0.5) 0.14 -0.29
very low elasticities II (0.1, 0.05, 0.05, 0.5) 0.13 -0.26

Notes

This work has benefited from the support of the Agence Nationale de la Recherche through the program Investissements d’Avenir ANR-17-EURE-0001.

The contribution of Youssef Salib to this work has been completed partly during an internship at the CEPII and partly during the PhD program at PSE.

Bibliography

Caliendo, Lorenzo, and Fernando Parro. 2015. “Estimates of the Trade and Welfare Effects of NAFTA.” The Review of Economic Studies 82 (1): 1–44. https://doi.org/10.1093/restud/rdu035.
Chepeliev, M., and E. Corong. 2022. “Revisiting the Environmental Bias of Trade Policies Based on an Environmentally Extended GTAP MRIO Data Base.” Conference Paper. Center for Global Trade Analysis, Purdue University. GTAP.
Guerrero, Santiago, Ben Henderson, Hugo Valin, Charlotte Janssens, Petr Havlik, and Amanda Palazzo. 2022. “The Impacts of Agricultural Trade and Support Policy Reform on Climate Change Adaptation and Environmental Performance: A Model-Based Analysis.” OECD Food, Agriculture and Fisheries Papers 180. OECD Publishing. https://doi.org/10.1787/520dd70d-en.
Klotz, Richard, and Rishi R. Sharma. 2023. “Trade Barriers and CO\(_2\).” Journal of International Economics 141 (March): 103726. https://doi.org/10.1016/j.jinteco.2023.103726.
Laborde, David, Abdullah Mamun, Will Martin, Valeria Piñeiro, and Rob Vos. 2020. “Modeling the Impacts of Agricultural Support Policies on Emissions from Agriculture.” IFPRI; National Bureau of Economic Research. https://doi.org/10.3386/w27202.
Moreira, Mauricio Mesquita, and Marcelo Dolabella. 2024. “Does Trade Policy Help or Hinder Global Warming? A Case Study of Latin America and the Caribbean.” The World Economy 47 (2): 779–805. https://doi.org/10.1111/twec.13449.
Shapiro, Joseph S. 2021. “The Environmental Bias of Trade Policy.” The Quarterly Journal of Economics 136 (2): 831–86. https://doi.org/10.1093/qje/qjaa042.