Stranded fossil-fuel assets translate to major losses for investors in advanced economies

The global transition to a low‑carbon economy entails deep and rapid structural change, including the accelerated phase‑out of fossil‑fuel production. This creates transition risk by forcing premature retirement or write‑downs of functioning capital assets and reserves. Although over 100 studies have examined scenario‑contingent early retirement of fossil‑energy supply facilities, prior work has not linked asset‑level stranding to financial ownership, leaving the distribution of financial exposure insufficiently understood. Asset stranding is defined as the collapse of expectations of future profits from invested capital due to disruptive policy and/or technological change, which is reflected in enterprise value and market prices and can transmit through financial networks. These effects can raise social concerns by destabilizing financial markets with repercussions for the real economy, including pensions and public finances, especially given the scale of change needed to meet 1.5–2°C goals. This paper maps the global financial geography of stranded oil and gas asset risk along equity ownership chains from extraction sites to ultimate owners, distinguishing geographic and functional characteristics, to reveal who ultimately bears wealth losses and how risks propagate.

Prior literature has emphasized that large amounts of fossil‑fuel reserves will remain unused to meet climate goals and has analyzed transition and systemic financial risks from climate scenarios. Regulatory climate stress tests typically apply synthetic shocks to financial assets without tracing them from underlying real assets. Studies also document that non‑bank financial institutions, notably funds and pensions, can be heavily exposed to climate risks, and that sector‑wide assumptions of uniform risk can be misleading due to heterogeneity in costs, ownership, and behavior. This study builds on and connects asset‑level stranded asset analyses with financial network ownership tracing, addressing the gap in linking physical asset stranding to ultimate equity owners and the distribution of losses across countries and institutions.

The study operationalizes asset stranding as the present value (PV) loss from a change in expectations regarding discounted future profit streams of upstream oil and gas assets. It uses Rystad Energy’s Ucube dataset covering 43,439 fields and 69,990 ownership shares, with asset‑level breakeven costs, reserves/resources, and owners of record (3,113 companies). Profit losses are computed as the discounted difference between baseline and policy scenario profits over a 15‑year horizon, assuming expectations realign in 2022 and using a 6% annual discount rate. Scenarios for demand and prices are generated by the E3ME‑FTT‑GENIE integrated assessment framework, coupling a macroeconometric model (E3ME), an energy technology diffusion model (FTT), and a carbon cycle–climate model (GENIE). The baseline represents investor expectations (IEA WEO 2019 Current Policies, ~3.5°C), and the policy scenario (EUEA) reflects EU and East Asia net‑zero pledges (~2°C), including sell‑off behavior where low‑cost Middle Eastern producers increase market share as demand declines. Alternative realignments (benign, severe, quota, and sensitivity to CCS availability and network imputations) are also examined. Asset‑level stranded profits are aggregated to four ownership stages: (1) physical location of fields; (2) fossil‑fuel company headquarters; (3) ultimate corporate owners via an equity ownership network; and (4) ultimate owners (governments, individuals/fund investors, creditors, and unknown). Ownership tracing uses Bureau van Dijk ORBIS data curated via a snowballing download to build a global network of 1,772,899 company nodes and 3,196,429 equity links. A network propagation algorithm transmits shocks from fossil‑fuel companies to their owners proportional to equity stakes, with managed fund holdings tracked but not re‑propagated. Balance sheets are adjusted by subtracting shocks from assets and equity; propagation halts at technical insolvency (shock exceeding equity), with excess recorded as creditor losses. Ownership loops are identified and bypassed to ensure convergence. Missing financials (equity and assets) for companies are imputed using regressions on available variables (assets, equity, revenue, employees, size), applied stochastically while enforcing assets ≥ equity. Results are reported under consolidated accounting (weighting shocks by the fraction not owned by other companies) for stages 3–4, with an alternate unweighted tally to bound listed‑company and financial‑sector exposures. Robustness checks include alternative realignments, removal of CCS, and network imputation sensitivity.

- Total stranded assets: In the medium realignment (InvE_EUEA-SO), present value of upstream oil and gas lost profits equals about US$1.4 trillion over 15 years (discounted at 6%). - Geographic propagation to OECD: Stage 1 (physical fields) places 39.2% (US$552 billion) in OECD countries; Stage 2 (headquarters) rises to 51.7% (US$728 billion); Stage 3 (ultimate corporate owners) peaks at 57.1%; Stage 4 redistributes about 1.6% back to non‑OECD via non‑OECD clients of OECD asset managers. Net international transfer of more than 15% of global stranded asset risk accrues to OECD investors; result is robust across scenarios. - Institutional breakdown: Roughly US$1.0 trillion of losses are booked by listed oil and gas companies at stage 2. At stage 3, the financial sector bears US$438 billion of losses (about 88% in OECD). At stage 4, governments directly own about US$484 billion (34%), with the remainder mostly borne by private individuals (including via funds). Losses exceed equity by US$129 billion in 239 companies (total debt US$361 billion), implying technical insolvencies. - Country patterns: Physical stranding is largest in the United States and Russia (~US$300 billion each), followed by China and Canada (~US$100 billion each). Low‑cost Middle Eastern producers (Qatar, Saudi Arabia, Iran) have comparatively modest losses (<US$50 billion) due to continued profitability and sell‑off behavior. Financial risk is redistributed internationally: e.g., France imports most losses at stage 2; the United States receives large net transfers at stage 3 due to major asset managers; tax‑haven and financial‑center jurisdictions (e.g., British Virgin Islands, Switzerland) register notable transfers. - Listed companies and financial amplification: Listed companies cumulatively own about US$1.27 trillion of stranded assets, 19% of which are only revealed along the ownership chain. Summing potential losses at all financial institutions yields an upper bound of about US$681 billion potentially affecting the financial sector; up to ~US$400 billion could hit financial sector balance sheets (including via impaired collateral), a 29% amplification relative to direct stage‑2 losses. Funds carry a much larger share of risk than banks; about US$90 billion is owned directly by pension funds, with additional unknown exposures via asset managers. Losses in the US and UK financial sectors are especially large. - Robustness: The OECD net transfer (one‑sixth to one‑fifth of total losses) and the country ranking of ultimate losses are robust across scenario variants and sensitivity tests (including no‑CCS and no‑imputation cases).

By tracing ownership from physical assets to ultimate owners, the study shows that financial exposure to oil and gas stranding is more evenly spread geographically than production suggests, and substantially shifted to OECD private investors and financial markets. This has several implications. First, decarbonization policies in OECD countries can materially affect global supply because about half of at‑risk assets are operated by companies headquartered there (stage 2), countering the notion that supply risks are concentrated solely in producer countries. Second, ownership changes (e.g., divestments) can lead to ownership leakage rather than reduced systemic risk, as assets move to other owners capable of transmitting transition risk. Non‑bank financial institutions—particularly funds and pensions—are key transmission channels yet are less regulated than banks, increasing concerns about contagion. Third, at the ultimate owner level, losses fall on governments and individuals (including via funds); this may generate lobbying for public support, potentially shifting private losses to the public sector and creating moral hazard. Given that many exposures are in wealthy countries, bailouts may be fiscally feasible (often 1–2% of GDP under the medium realignment), which could affect investor expectations and investment behavior. The findings argue for stress tests and scenarios that incorporate heterogeneous risk distributions within sectors, recognition of international financial linkages, and coordinated policies to manage a stable phase‑out of fossil fuel production and reduce destabilizing realignments.

This paper comprehensively links asset‑level stranding of oil and gas production to the equity ownership chain, mapping losses from fields to corporate headquarters, financial intermediaries, and ultimate owners. It estimates roughly US$1.4 trillion in stranded assets under a medium realignment and shows a robust net transfer of more than 15% of global stranded asset risk to OECD investors, with significant exposures among listed companies, funds, pensions, and financial centers. The results highlight the political economy stakes for rich‑country stakeholders, the potential for financial amplification, and the importance of regulating and monitoring non‑bank financial institutions. Policy implications include: integrating ownership tracing into climate‑related financial risk assessments; designing stress tests that account for heterogeneous exposures; supporting international cooperation to manage an orderly fossil‑fuel phase‑out; and aligning financial sector incentives (e.g., via green taxonomies and stewardship) to reduce ownership leakage and limit future stranded asset creation. Future research could extend to debt‑channel contagion, dynamic feedbacks from asset sales and fire‑sales, household‑level wealth distribution effects, and sectoral spillovers in macro‑financial models.

Key limitations include: reliance on proprietary datasets (Rystad Ucube, ORBIS) and manual matching, which may omit or misclassify some entities; missing financial data imputed stochastically for many companies, though aggregate results proved robust; exclusion of debt‑network distress and fire‑sale dynamics (equity channel only), implying that total systemic impacts may be underestimated; technical treatment of ownership loops by bypassing identified circular links; partial information on ultimate owners (some losses assigned to ‘unknown’) and potential foreign ownership not fully captured; scenario assumptions (timing of expectations realignment in 2022, 15‑year horizon, 6% discount rate, policy pathways including sell‑off behavior and CCS availability) which affect magnitudes but not the main qualitative findings; and consolidated accounting choices that bound listed and financial‑sector loss estimates between aggregated and fully disaggregated views.