Allocating capital-associated CO₂ emissions along the full lifespan of capital investments helps diffuse emission responsibility

The study addresses how to assign CO₂ emission responsibilities associated with capital assets (machinery, infrastructure) whose lifespans span years to decades. Because capital is produced by specific sectors (capital formation), used by other sectors for production (capital use), and persists across time, conventional accounting that assigns all embodied emissions to the formation year can misattribute responsibility. The research proposes allocating emissions embodied in capital over the full lifetime of assets to their users and ultimately to final demand over time. The purpose is to capture the temporal displacement of environmental responsibility, improve equity across generations, and better inform policy in economies with intense and rapid capital accumulation such as China, where capital investments constitute a large share of GDP and historically generated substantial CO₂ emissions.

Prior work has examined spatial displacement of environmental pressures along supply chains, moving from producer- to consumer-based accounting. Conventional analyses treat capital assets like final consumption goods, assigning production emissions to purchasing sectors or regions via gross fixed capital formation (GFCF). More recent approaches endogenize capital as a production factor in MRIO models, allocating capital-associated pressures to final consumption across sectors and countries. Chen et al. incorporated capital dynamics for a single year, but intertemporal features—capital cohorts produced under different technologies and emissions intensities—remain largely unaddressed. Neglecting these temporal aspects can underestimate capital-associated GHG emissions by about 30%. The study situates itself as extending capital-endogenized MRIO modeling by explicitly tracing capital flows across sectors/regions and along lifespans, reallocating embodied emissions from past investments to current and future production and consumption.

The authors build an inter-provincial, capital-endogenized multi-regional input-output (MRIO) model for China to trace and allocate capital flows and embodied CO₂ over 1995–2017 and under scenarios to 2030.

• Capital data: Provincial sectoral capital investment time series are constructed primarily from Newly Increased Fixed Assets (NIFA) statistics, adjusted upward for underreporting using national adjustment factors (λ) following Wu. Investments are disaggregated into four asset categories (equipment, residential structures, non-residential structures, others; with 'others' reallocated 3:7 to equipment and non-residential structures) using available TIFA subcategory data and industrial bulletins. The perpetual inventory method (PIM) is applied to compute capital consumption (depreciation) time series by sector, asset, province.
• Capital endogenization: Capital consumption (depreciation) flows D(t≤n) are mapped between capital-producing and capital-using sectors via concordance from the 37-sector capital dataset to 42-sector MRIO classification, yielding a matrix where each element records assets invested in year t and depreciated in year n. Monetary values are expressed in 2017 USD using exchange rates and CPI.
• MRIO tables: Inter-provincial MRIO series are compiled using benchmark tables (1995–2017) from CEADS and Wang, adjusted to align with national statistics (final demand, exports, imports, value-added), and rebalanced using the GRAS method for missing years.
• Emissions reallocation: Supply chain-wide CO₂ embodied in capital depreciation in year n but emitted in year t is computed as F_t = S_t L_n D_t→n, where S_t are direct CO₂ intensities (from CEADS), L_n is the Leontief inverse for year n. F_t is allocated either to production of capital-using sectors (yielding production-based emissions after reallocation, PBE^k) or to final demand categories (final consumption, GFCF, exports) to obtain consumption-based emissions after reallocation (CBE^k). Conventional PBE and CBE related to GFCF are removed and replaced with reallocated F across years to capture temporal assignment.
• Scenarios to 2030: Three pathways are modeled by manipulating the 2017 MRIO and projecting to 2030: • BAU: Continuation of historical trends in population, efficiency, productivity; GDP growth 6.5%/yr pre-2020, 5%/yr post-2020. Efficiency improvements reduce intermediate inputs; energy mix adjusted per IEA; exports scale with GDP; tables rebalanced via GRAS; emissions adjusted with sectoral intensity changes. • KES (capital for economy and social well-being): Additional infrastructure-focused investments (roads, rail, airports, seaports, electricity/water supply, telecommunications) layered on BAU, allocated to investing sectors and capital-producing sectors per observed shares; induced changes in related final consumption; MRIO and emissions adjusted. • KLC (capital for low-carbon development): Targeted investments in low-carbon technologies in power and end-use sectors (e.g., CCS, EVs) based on IEA World Energy Outlook 2017. Energy mix shifts toward low-carbon, exports decrease and imports increase to reduce territorial emissions; sectoral direct CO₂ intensities adjusted accordingly; MRIO rebalanced. Uncertainty analyses report interquartile ranges.
• Definitions: The study introduces 'historically committed CO₂ emissions'—emissions embodied in pre-year-n capital that are allocated to production/consumption in year n via depreciation—complementing forward-looking 'committed emissions' from operating existing fossil infrastructure.

• Temporal reallocation effect: Allocating emissions embodied in capital over asset lifespans substantially reduces responsibility attributed to the formation year. National PBE^k is 25–35% lower than conventional PBE since 1995; national CBE^k is 31–46% lower than conventional CBE. Only about one-third of CO₂ embodied in 1995–2017 GFCF is assigned to production within the same period; the rest is allocated to 2018–2030 and post-2030 use.
• Misallocation under conventional accounts: Conventional input-output estimates assign most 'capital investment' emissions to capital formation sectors (e.g., construction with 68% of CBE_GFCF in 1995–2017), rather than to actual capital users (e.g., real estate, transport, residential services). Reallocating to users and final demand across time materially changes sectoral and regional accounts.
• Per-capita and inequality impacts (2017): Regional per-capita PBE reductions of 15–38% relative to conventional PBE; larger PBE reductions in Northwest, North, Northeast (net exporters of capital and embodied CO₂). Per-capita CBE reductions larger in high-investment regions (e.g., Central and Southwest: −36%). Reallocation decreases interprovincial inequality in per-capita PBE, but slightly increases inequality in per-capita CBE.
• Scenario outcomes (2030): • BAU: National PBE increases by ~15% from 2017; CBE grows similarly, driven by materials and construction; electricity-sector emissions partially offset by efficiency and energy mix shifts. • KES: PBE increases by ~20% (largest uncertainty −4% to +6% around projections); boosted infrastructure investment raises emissions and GDP. • KLC: Modest emissions growth (overall <2% vs 2017 in the narrative); an extra 7% investment in low-carbon technologies yields ~9% reduction in national PBE compared with BAU but with ~4% GDP decrease.
• Historically committed share: About 10% of national CO₂ emissions in 2030 (PBE or CBE) stem from CO₂ embodied in 1995–2017 capital. In 2018–2019 this share is higher (23–30%). Total capital-associated shares (historical + 2018–2030) reach 32–34% of PBE and 37–39% of CBE in 2030 (±2% uncertainty).
• Sectoral composition of committed emissions (2030): Electricity generation is dominated by future committed emissions, with historically committed shares only ~4–6%. Service sectors are dominated by historically committed emissions: real estate and residential services exceed 83% historically committed (both PBE and CBE). Transportation services have >60% future committed shares. Under KLC, cleaner future production increases the relative share of historically committed emissions.
• Policy-relevant example: Using the time-allocation scheme would reduce energy-sector PBE by about 30% relative to conventional in 2030 across scenarios, implying implications for emissions trading allocation methods.

The findings demonstrate that endogenizing capital and allocating embodied emissions over asset lifetimes shifts responsibility from capital producers to capital users and diffuses emission responsibility across time. This addresses the core research question by capturing temporal displacement inherent in durable capital and improving intergenerational equity in accounting. The introduced concept of 'historically committed CO₂ emissions' provides a backward-looking complement to forward-looking 'committed emissions', offering a comprehensive view of how past investment decisions affect current and future emission accounts. The over-time accounting scheme resembles a 'mortgage' of emission burdens—formation-year emissions are virtually spread over the payback period—highlighting both the risk of delayed mitigation and a framework for aligning with carbon-neutrality pathways if paired with offsetting measures. Results emphasize that service sectors bear substantial historically committed burdens, challenging perceptions that they are low emitters when capital is accounted for. Scenario analysis indicates that targeted low-carbon capital investment (KLC) can deliver national-scale emissions reductions, albeit with GDP trade-offs, and that earlier deployment of efficient capital reduces future emissions intensity. The work suggests implications for policy instruments such as emissions trading, where accounting choices materially affect allowances, and outlines data and methodological needs to integrate capital-associated emissions into regulatory frameworks.

This study develops an inter-provincial capital-endogenized MRIO framework for China (1995–2017) and scenario analysis to 2030 to allocate capital-associated CO₂ emissions over the full lifespans of assets. It shows that temporal allocation reduces current-year emission responsibility by 25–46% relative to conventional accounts, reveals significant historically committed shares persisting into 2030 (∼10% of national emissions), and reattributes responsibility from capital producers to users and final demand. Sectoral and regional analyses uncover substantial historically committed burdens in capital-intensive services and highlight how low-carbon investment pathways can reduce future emissions. The introduced 'historically committed' metric offers a practical scheme for intertemporal responsibility assignment and supports more equitable, forward-looking policy design. Future research should develop higher-resolution capital asset datasets, improve valuation of capital services versus consumption, integrate dynamic economic models with MRIO to capture feedbacks, and explore alternative capital development narratives and their economy–energy–emissions interactions.

• Data uncertainties from multiple sources and adjustments (e.g., underreporting correction λ, asset disaggregation) affect precision; early-period estimates are conservative due to omission of pre-1995 capital.
• Use of capital consumption (depreciation) rather than capital services is debated; capital services require asset price data with high uncertainty.
• Limited asset categorization increases uncertainty in depreciation/emission properties; IO aggregation issues persist when heterogeneous products are combined into sectors.
• Static MRIO framework cannot capture dynamic feedbacks between investment and consumption across sectors; linear correlations between investment and consumption changes are assumed in scenarios.
• Scenario assumptions (GDP growth, energy mix, export/import responses) and mapping from IEA projections introduce uncertainty; results indicate relative changes and trends rather than precise forecasts.
• Potential methodological implications for policy tools (e.g., ETS) need dedicated, high-resolution datasets (asset lifetimes, plant-level details) and standardized accounting methods to operationalize historically committed emissions.