## 1. Altruistic Optimization

## Source details

**Canonical URL:** [1. Altruistic Optimization](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021280-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021280-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021280-print-pdf.pdf.json)

---

### I. Introduction
- Climate financing commitments and estimates:
  - Advanced economies would provide US$100 billion annually for low-income countries (terms being reworked in the context of COP26).
  - OECD (2020) estimates that climate financing was close to US$80 billion in 2018, with a grant element of about 10 percent.
- Two policy dimensions:
  - Purpose: adaptation versus mitigation.
  - Shortcoming addressed: narrow financing (liquidity need) versus compensation (transfer element).
- Paper objectives:
  - Analyze links between development, climate mitigation policies, and compensation for EMDEs in a sequential two-block development model (AEs develop first, EMDEs second).
  - Key questions:
    - Implications of sequential development pattern for mitigation incentives in AEs and EMDEs.
    - Circumstances for an equity case for compensating EMDEs.
    - Scope for mutually beneficial compensation (efficiency case) where AEs transfer wealth to induce EMDEs to abandon fossil-fuel development.
    - Role of relative size and asymmetric climate vulnerability.
- Second part of paper (section V) adapts model to incorporate diffusion of GHG-free energy technologies and examines effects on development patterns, climate policies, and compensation.

### II. The Basic Model
- Development and emissions:
  - Production capacity P built up over time is a function of cumulative GHG emissions e_i (a stock and irreversible).
  - Net development benefit for country i: P_i(e_i) − C_i(E), where E = ∑ e_j.
- Core equations and technical conditions:
  - (1) P_i = f(e_i),    f′(e_i) > 0, f′′(e_i) < 0
  - (2) C_i = g(E),    g′(E) > 0, g′′(E) > 0,    E = ∑ e_j
  - Boundary conditions: f(0) = g(0) = 0 and f′(0) > g′(0)
- Parameterization for illustration:
  - f(e_i) = e_i^α with 0 < α < 1
  - g(E) = E^β with β > 1
  - Illustrative parameter values: α = 0.5 and β = 2
- Game structure:
  - Two-stage game with two players: Advanced Economies (AEs) first, Emerging Markets and Developing Economies (EMDEs) second.
  - Reference model assumes AEs and EMDEs of equal size.
  - Sequential development simplifies collective-action problems by having only one group emit at a time.
- Empirical note preserved:
  - AEs accounted for almost all cumulative emissions until about 1950; since then EMDE share rose to about 35 percent today.

### Key implications from the sequential-development game
- Emission saturation and mitigation incentives:
  - AEs emit until a saturation point e̅_AE where f′(e̅_AE) = g′(e̅_AE); at that point AEs have incentive to implement mitigation.
- EMDE incentives and timing:
  - EMDEs begin development after AEs’ saturation point; high returns at early stages make fossil-fuel driven development profitable for EMDEs until their own saturation point defined by f′(e̅_EMDE_A) = g′(e̅_AE + e̅_EMDE_A).
  - EMDEs will forgo emissions only once their private marginal benefit equals the global marginal cost given existing AE emissions.
- Distributional and strategic features:
  - AEs prefer EMDEs to mitigate but EMDEs are unwilling because mitigation would forgo profitable development opportunities.
  - EMDE development increases global GHG emissions and costs, harming AEs without generating additional AE benefits.
  - When global emissions finally end (EMDE saturation), AEs have accumulated higher production capacity than EMDEs.
- Strategic considerations:
  - Baseline assumes AEs optimize individually; strategic optimization (anticipating EMDE responses) is theoretically appealing but potentially unrealistic because of long horizons and uncertainty.

### Climate compensation: equity versus efficiency
- Equity case:
  - AEs would need to transfer an amount reimbursing EMDEs for costs AEs caused by using up atmospheric absorptive capacity to equalize net benefits.
  - Equity case is unconditional but weakens if EMDEs are large enough to impose large costs on AEs.
- Efficiency (Pareto-improvement) case and marginal condition:
  - Mutually profitable transfer possible when AEs’ marginal cost of one additional unit of GHG exceeds EMDEs’ marginal private benefit:
    - (3) g′(e̅_AE + e_EMDE_A) (A) > f′(e_EMDE_A) − g′(e̅_AE + e_EMDE_A) (B) > 0.
    - (A) is the cost of one additional unit of GHG emissions for AEs.
    - (B) is the net benefit of one additional unit of GHGs for EMDEs.
  - If (A) > (B), AEs can offer an amount between (B) and (A) to buy EMDEs out of emissions.
- Key results:
  - Under model assumptions, a mutually beneficial compensation scheme always exists because EMDEs obtain near-zero net benefit from their last unit while that unit imposes non-zero cost on AEs.
  - EMDEs will exploit the most profitable opportunities and expand emissions beyond AEs’ saturation before agreeing to be bought out.
  - Negotiated compensation outcomes lie between AEs’ maximum willingness to pay and EMDEs’ minimum required compensation; negotiated rent is indeterminate ex ante.

### Variations: relaxing symmetry assumptions
- A. EMDEs are Large
  - If EMDEs are larger (illustrated by tripling EMDEs’ size), EMDE development curve shifts outward.
  - Some baseline results persist: relative to size, AEs build up more capacity and emit more GHGs than EMDEs; AEs would still buy EMDEs out of some emissions.
  - Key difference: AEs’ first-developer advantage can unravel as large EMDEs reach higher saturation points, increasing costs imposed on AEs and weakening the equity case for compensation.
- B. EMDEs disproportionately vulnerable to climate change
  - Doubling the cost from global warming for EMDEs relative to AEs reduces EMDEs’ room for fossil-fuel driven development.
  - This amplifies AEs’ first-developer advantage and strengthens the equity case for compensation.
  - However, climate-vulnerable EMDEs are in a weak negotiation position because they would likely mitigate without compensation—so buy-out schemes would be small.
  - Heterogeneity implies divergent interests: smaller, climate-vulnerable EMDEs may align with AEs and accept modest compensation; large temperate EMDEs would be costlier to buy out.
- Empirical population detail preserved:
  - EMDEs host a far larger share of the world’s population than advanced economies: 86 percent.

### Energy efficiency, clean energy technologies, and technology transfers
- Energy efficiency:
  - Improves returns to energy use, shifting development curve outward and potentially raising both production capacity and GHG emissions if energy remains fossil-fuel based.
  - In the model, raising energy efficiency is formally identical to increasing an economy’s size.
- Clean energy technologies:
  - Allow higher production capacity without net increase in GHG emissions; modeled as a cost function with lower marginal cost than continued GHG emissions at AEs’ saturation point.
  - Availability to EMDEs is a positive externality (technology transfer).
- Implications of adoption:
  - Saturation points shift right for both AEs and EMDEs; both can build more production capacity without increasing GHGs.
  - No efficiency case for compensation remains: EMDEs would stop emitting GHGs by themselves upon adoption of clean technologies.
  - First-developer equity advantage persists but shrinks as marginal cost of clean energy falls.
  - Technology transfer is superior to financial compensation: it reduces GHGs and raises net development benefits for both AEs and EMDEs.
- Policy-relevant qualifications:
  - Technology transfer encompasses reducing trade barriers, resolving intellectual property issues, financing for low-income countries, advisory initiatives, and ambitious mitigation measures in AEs (credible targets, carbon pricing, R&D support).
  - Both financial and technology transfers may be needed in practice because technological progress is uneven across sectors.

### Summary and policy conclusions
- Divergent interests justify differentiated mitigation by development needs: AEs may face a point where additional GHG emissions cost more than growth benefits while some EMDEs still find fossil-fuel development beneficial.
- Climate compensation:
  - Equity case: motivated by AEs’ prior use of atmospheric absorptive capacity; strongest when EMDEs are modestly sized relative to AEs and when EMDEs are climate-vulnerable; can be unconditional or linked to adaptation.
  - Efficiency case: exists when EMDEs’ continued emissions impose larger damages on AEs than EMDEs’ private benefits; stronger when EMDEs are large relative to AEs; weaker when EMDEs are climate-vulnerable or when clean technologies are available.
  - Equity and efficiency cases differ in motivation and applicable circumstances and should be treated separately in policy design.
- Technological development and transfer:
  - Development and transfer of clean energy technologies is ultimately superior to financial compensation, dissolving the climate-development conundrum and improving growth prospects for both AEs and EMDEs.
  - AEs have strong incentives to invest in clean technologies as they approach fossil-fuel saturation points and to ensure EMDEs can use them.
- Policy implication:
  - AEs pursuing mitigation policies is in their own interest even if EMDEs do not immediately reciprocate, because such policies incentivize development and diffusion of clean technologies.

### Annex: comparison of individual, altruistic and strategic optimization
- Behavioral scenarios for AEs:
  - Individual optimization (baseline):
    - First-order condition (ind): (A1) f′(e̅_AE_ind) = g′(e̅_AE_ind)
    - Interpretation: AEs stop emitting when their marginal development benefit equals their marginal external cost.
  - Altruistic optimization:
    - AEs maximize joint net development benefit (AEs + EMDEs).
    - Altruistic first-order conditions and symmetry:
      - (A4) f′_alt(E_alt) = g′_alt(E_alt)
      - (A5) e̅_AE_alt = e̅_EMDE_A_alt = 1/2 E_alt (size symmetry)
    - Comparative statics:
      - (A6) f′(1/2 E_alt) > g′(1/2 E_alt)
      - Implies (A7) e̅_AE_individual > e̅_AE_alt
      - From (A3) and (A7): (A8) e̅_AE_individual + e̅_EMDE_A_individual > e̅_AE_alt + e̅_EMDE_A_alt
      - Conclusion: Individual optimization results in higher overall GHG emissions than altruistic optimization.
  - Strategic optimization (Stackelberg leader):
    - AEs anticipate their emissions' effect on EMDE emissions:
      - (A9) C_AE = g(e_AE + e_EMDE_A(e_AE))
      - First-order condition:
        - (A10) f′(e̅_AE_strat) = g′(e̅_AE_strat + e̅_EMDE_A_strat) × [1 + (∂e̅_EMDE_A_strat / ∂e̅_AE_strat)]
    - Comparative implications and ambiguity:
      - Induced EMDE emissions are smaller than AE emission increases (from A3), but occur at higher atmospheric GHG concentration and are costlier per unit.
      - Net effect of strategic behavior versus individual optimization is ambiguous and parameter-dependent.
      - For the base specification in the main text, differences between individual and strategic optimization are minimal.

*wpiea2021280-print-pdf - 1. Altruistic Optimization*

### 1. Altruistic Optimization  _____________________________________________ 17

### 1. Altruistic Optimization

### I. Introduction
- Climate financing is playing an increasingly prominent role in the international climate debate and is enshrined in the Paris climate accord; advanced economies would provide US$100 billion annually for low-income countries (terms being reworked in the context of COP26).
- OECD (2020) estimates that climate financing was close to US$80 billion in 2018, with a grant element of about 10 percent.
- Two dimensions of climate financing discussed:
  - Purpose: adaptation (contain economic and social repercussions) versus mitigation (reduce/offset greenhouse gas emissions to limit global warming).
  - Shortcoming addressed:
    - Narrow financing (liquidity need): e.g., thin and incomplete insurance markets for climate risk; most relevant for adaptation.
    - Compensation (transfer element): reimburses lower-income countries for damage inflicted by GHG emissions of advanced economies; often conditional on policy objectives and delivered via grants, debt relief, or concessional lending with grant elements.
- Paper’s objectives and questions:
  - Analyze links between development, climate change mitigation policies, and compensation for EMDEs in a sequential two-block development model (AEs develop first, EMDEs develop second).
  - Key questions the model addresses:
    - What does the sequential development pattern imply for climate change mitigation? Are mitigation policies desirable for both AEs and EMDEs?
    - Under which circumstances is there an equity case for compensating EMDEs?
    - Is there an efficiency case for compensation — scope for mutually beneficial schemes where AEs transfer wealth to EMDEs in return for EMDEs abandoning fossil-fuel driven development?
    - How do relative size of AEs vs. EMDEs and asymmetric climate vulnerability affect mitigation policies and equity/efficiency cases for compensation?
- Second part of paper (section V) adapts the model to incorporate diffusion of GHG-free energy technologies and examines effects on development patterns, climate policies, and compensation.

### II. The Basic Model
- Development pattern: built-up of production capacity involves burning fossil fuels and emitting greenhouse gases (GHGs).
- Model equations and assumptions (as presented):
  - (1) 푃푃푖 = 푓푓(푒푒푖),    푓푓′(푒푒푖) > 0, 푓푓′′(푒푒푖) < 0
    - captures an environmental Kuznets-type relationship: production capacity P built up over time is a function of cumulative GHG emissions 푒푒푖 (a stock and irreversible).
  - (2) 퐶퐶푖 = 푔푔(퐸퐸),    푔푔′(퐸퐸) > 0, 푔푔′′(퐸퐸) > 0, 퐸퐸 = ∑ 푒푒푗
    - cost function is nonlinear in total global emissions E = sum of all countries’ emissions; costs negligible at low emissions, rising disproportionately with higher concentrations.
  - Technical conditions to ensure equilibrium with positive GHG emissions:
    - 푓푓(0) = 푔푔(0) = 0
    - 푓푓′(0) > 푔푔′(0)
- Parameterization used for illustration:
  - 푓푓(푒푒푖) = 푒푒푖^훼 with 0 < 훼 < 1
  - 푔푔(퐸퐸) = 퐸퐸^훽 with 훽 > 1
  - Illustrative parameter values: 훼 = 0.5 and 훽 = 2 (most results do not depend on these precise values provided the non-linearity conditions hold).
- Model structure:
  - Two-stage game with two players: Advanced Economies (AEs) develop first; Emerging Markets and Developing Economies (EMDEs) develop second.
  - Reference model assumes AEs and EMDEs are of equal size (later modified).
  - Simplification: only one country group emits at any given time due to sequential development, eliminating collective action problems in the basic formulation.
  - Empirical note: AEs accounted for almost all cumulative emissions until about 1950; since then EMDE share rose to about 35 percent today.

### Key implications from the sequential-development game (as described)
- Emission saturation and mitigation incentives:
  - As AEs develop, they emit GHGs until emissions reach a saturation point 푒푒̅ 퐴퐴퐴퐴 where marginal benefit equals marginal cost: 푓푓′(푒푒̅ 퐴퐴퐴퐴) = 푔푔′(푒푒̅ 퐴퐴퐴퐴).
  - At that point f ossil-fuel driven development in AEs ends; AEs have an incentive to implement mitigation policies that arrest additional emissions.
- EMDEs’ incentives differ:
  - EMDEs begin development after AEs’ saturation point. Returns from GHG emissions are high at early development stages, so EMDEs find profitable opportunities in fossil-fuel driven development despite increased global costs.
  - EMDEs will only abandon fossil-fuel driven development once they reach their own saturation point defined by 푓푓′(푒푒̅ 퐴퐴퐸퐸퐸퐸  퐴퐴) = 푔푔′(푒푒̅ 퐴퐴퐴퐴 + 푒푒̅ 퐴퐴퐸퐸퐸퐸  퐴퐴).
- Distributional and strategic features:
  - At AEs’ saturation, AEs will implement mitigation; they would prefer EMDEs do the same, but EMDEs will be unwilling because mitigation would forgo profitable development.
  - EMDE development is harmful to AEs: it increases global GHG emissions and their associated costs without generating additional benefits for AEs.
  - When global emissions finally end (EMDE saturation point), AEs will have accumulated higher production capacity than EMDEs.
- Strategic vs. individual optimization:
  - The model assumes AEs optimize individually (maximize own net benefits) and do not endogenize EMDE behavior; strategic optimization (anticipating EMDE responses) is discussed as theoretically appealing but potentially unrealistic given long time horizons and uncertainty.

*Italic source attribution: wpiea2021280-print-pdf - 1. Altruistic Optimization*

### Annex for a  comparison of outcomes under individual, altruistic and strategic optimization.

### Annex for a  comparison of outcomes under individual, altruistic and strategic optimization.

### Model framework and key mechanism
- Two-player, two-stage game: advanced economies (AEs) develop first (first-mover advantage), then emerging markets and developing economies (EMDEs) develop.
- Development driven by fossil-fuel use emits irreversible GHGs that reduce atmospheric absorptive capacity, imposing global warming costs equally on all.
- Net development benefit (for country i) is 푃푃_푖(푒푖) − 퐶퐶_푖(𝐸), where development capacity gains are offset by global-warming costs.
- Symmetric baseline: AEs and EMDEs have same size, same vulnerability, and identical relationship between GHG emissions and development, except for AEs’ sequential first-developer advantage.
- First-developer advantage arises because AEs use up the atmosphere’s absorptive capacity during early fossil-fuel driven development, leaving less capacity for EMDEs.

### Climate compensation: equity versus efficiency
- Equity case:
  - AEs would need to transfer an amount to EMDEs that reimburses EMDEs for costs AEs caused by using up atmospheric absorptive capacity to equalize net benefits.
  - Equity case is unconditional but weakens if EMDEs are large enough to impose large costs on AEs themselves.
- Efficiency (Pareto-improvement) case:
  - Possible if a transfer from AEs to EMDEs to induce EMDEs to abandon fossil-fuel development yields mutual gains.
  - Marginal-condition for a mutually profitable transfer (equation labeling preserved):
    - (3) 푔′(푒̅_퐴퐸퐸퐴 + 푒_퐴퐸퐸퐸_퐴) (퐴) > 푓′(푒_퐴퐸퐸퐸_퐴) − 푔′(푒̅_퐴퐴 + 푒_퐴퐸퐸퐸_퐴) (퐵) > 0.
    - (A) is the cost of one additional unit of GHG emissions for AEs.
    - (B) is the net benefit of one additional unit of GHGs for EMDEs.
  - If (A) exceeds (B), AEs can offer an amount larger than (B) but smaller than (A) to buy EMDEs out of emissions.
- Key results about compensation schemes:
  - A mutually beneficial compensation scheme always exists under the model’s assumptions because EMDEs obtain (by definition) near-zero net benefit from their last unit of GHGs while that unit imposes non-zero cost on AEs.
  - EMDEs will exploit the most profitable development opportunities and expand GHG emissions beyond AEs’ saturation point before agreeing to be bought out of remaining emissions.
  - Exact net development benefits under compensation cannot be determined ex-ante; negotiated outcomes lie between AEs’ maximum willingness to pay and EMDEs’ minimum required compensation, with the difference being a negotiable rent.

### Variations: relaxing symmetry assumptions
- A. EMDEs are Large
  - If EMDEs are larger relative to AEs (example illustrated by tripling EMDEs’ size), EMDEs’ development curve shifts outward.
  - Some baseline results persist: relative to size, AEs build up more capacity and emit more GHGs than EMDEs; AEs would still buy EMDEs out of some emissions.
  - Key difference: AEs’ first-developer advantage can unravel as large EMDEs exploit more opportunities and reach their saturation point at higher GHG levels, increasing costs imposed on AEs and weakening the equity case for compensation.
- B. EMDEs disproportionately vulnerable to climate change
  - Doubling the cost from global warming for EMDEs relative to AEs reduces EMDEs’ room for fossil-fuel driven development (cost overwhelms benefits).
  - This amplifies AEs’ first-developer advantage and strengthens the equity case for compensation.
  - However, climate-vulnerable EMDEs are in a weak negotiation position because they would likely take significant mitigation measures without compensation—so any buy-out scheme would be small.
  - Heterogeneity across EMDEs implies divergent interests: smaller, climate-vulnerable EMDEs may align with AEs and accept modest compensation; large temperate EMDEs would be costlier to buy out.

- Numeric/empirical detail preserved from source:
  - EMDEs host a far larger share of the world’s population than advanced economies: 86 percent (as cited from the IMF’s World Economic Outlook).

### Energy efficiency, clean energy technologies, and technology transfers
- Energy efficiency improvements:
  - Increase returns to energy use, shifting the development curve outward and ultimately raising both production capacity and GHG emissions if energy remains fossil-fuel based.
  - In the model, raising energy efficiency is formally identical to increasing an economy’s size.
- Clean energy technologies:
  - Represent technologies that allow higher production capacity without a net increase in GHG emissions.
  - Modeled as a cost function with lower marginal cost than continued GHG emissions at AEs’ saturation point.
  - Availability to EMDEs constitutes a positive externality (technology transfer); the technology does not produce negative externalities like GHGs.
- Implications of clean technology adoption:
  - Saturation points shift right for both AEs and EMDEs; both can build more production capacity without increasing GHGs.
  - No efficiency case for compensation remains: EMDEs would stop emitting GHGs by themselves upon adoption of clean technologies.
  - First-developer equity advantage persists because AEs previously passed part of the cost of an energy-intensive early development phase onto EMDEs via atmospheric emissions; however, the advantage shrinks as marginal cost of clean energy falls.
  - Technology transfer is superior to financial compensation: it reduces GHGs and raises net development benefits for both AEs and EMDEs.
- Policy-relevant qualifications:
  - Technology transfer is less precisely defined than financial transfers and encompasses a range of policies: reducing trade barriers, resolving intellectual property issues, financing for low-income countries, advisory initiatives, and importantly ambitious mitigation measures in AEs to create demand for low-carbon technologies (credible emission targets, carbon pricing, support for R&D).
  - Both financial and technology transfers may be needed in practice because technological progress is uneven across sectors (e.g., more advanced in power generation and industry than in long-distance transport or buildings).

### Summary and policy conclusions
- Divergent interests: AEs may face a point where additional GHG emissions cost more than growth benefits, while some EMDEs may still find fossil-fuel driven development beneficial—supporting differentiated mitigation requirements by development needs.
- Climate compensation:
  - Equity case: motivated by AEs’ prior use of atmospheric absorptive capacity; strongest when EMDEs are modestly sized relative to AEs and when EMDEs are climate-vulnerable (unconditional or linked to adaptation measures).
  - Efficiency case: can exist when EMDEs’ continued emissions impose larger damages on AEs than EMDEs’ private benefits; stronger when EMDEs are large relative to AEs; weaker when EMDEs are climate-vulnerable (they will mitigate on their own) or when clean technologies are available.
  - Equity and efficiency cases differ in motivation and applicable circumstances and should be treated separately in policy design.
- Technological development and transfer:
  - Development and transfer of clean energy technologies is ultimately superior to financial compensation, dissolving the climate-development conundrum and improving growth prospects for both AEs and EMDEs.
  - AEs have strong incentives to invest in clean technologies as they approach fossil-fuel saturation points and to ensure EMDEs can use them, thereby providing a global public good.
- Policy implication: AEs pursuing mitigation policies is in their own interest even if EMDEs do not immediately reciprocate, because such policies incentivize development and diffusion of clean technologies.

### Annex: altruistic and strategic optimization (behavioral scenarios for AEs)
- Baseline: independent optimization by AEs (AEs optimize independently of EMDEs):
  - First-order condition (independent optimization):
    - (A1) 푓′(푒̅_퐴퐴퐴퐴_푖푖푖푖푖푖) = 푔′(푒̅_퐴퐴퐴퐴_푖푖푖푖푖푖)
    - Interpretation: AEs stop emitting when their marginal development benefit equals their marginal external cost (ind).
- Alternatives sketched:
  - Altruistic optimization:
    - AEs maximize joint net development benefit (AEs + EMDEs).
    - Reveals biases from sequential, fossil-fuel driven development: over-development by AEs, under-development by EMDEs, and excessive GHG emissions overall.
  - Strategic optimization:
    - AEs maximize their net development benefit while anticipating EMDEs’ subsequent emissions behavior (Stackelberg leader).
    - Potentially more theoretically attractive than independent optimization but less realistic due to long horizons, policymaker time horizons, and deep uncertainty.
- EMDE behavior (common across scenarios):
  - EMDEs optimize development versus GHG emissions taking existing atmospheric GHGs as given:
    - (A2) 푓′(푒̅_퐴퐸퐸퐸_퐴) = 푔′(푒̅_퐴퐴 + 푒̅_퐴퐸퐸퐸_퐴)
  - Comparative implication:
    - (A3) −1 < (휕푒̅_퐴퐸퐸퐸_퐴 / 휕푒̅_퐴퐴퐴퐴) < 0: a higher saturation point for AEs implies a less-than-proportionate reduction in equilibrium GHG emissions by EMDEs, and therefore a less-than-proportionate increase in overall equilibrium emissions.

*Source: Annex for a comparison of outcomes under individual, altruistic and strategic optimization. (wpiea2021280-print-pdf)*

### 1.   Altruistic Optimization

### 1.   Altruistic Optimization

### Framework and first-order conditions
- (A4)     
  푓푓′_alt(퐸퐸_메_알트) = 푔푔′_alt(퐸퐸_메_알트)
- (A5)  푒̅_AAA_alt = 푒̅_AEEE_A_alt = 1/2 퐸퐸_메_알트
  - where alt denotes altruistic optimization and size symmetry implies AEs claim one-half of production capacity.

### Comparative static with separate (individual) optimization
- (A6)     
  푓푓′(1/2 퐸퐸_메_알트) > 푔푔′(1/2 퐸퐸_메_알트)
  - This inequality follows directly from the non-linearity assumptions (1) and (2) in the main text.
- Implication:
  - (A7) 푒̅_AAA_individual > 푒̅_AAA_alt
    - AEs optimizing independently use their first-mover advantage to emit a larger amount of GHGs than under altruistic behavior.
  - From (A3) and (A7) follows
    - (A8) 푒̅_AAA_individual + 푒̅_AEEE_A_individual > 푒̅_AAA_alt + 푒̅_AEEE_A_alt
  - Hence:
    - Individual optimization results in higher overall GHG emissions than altruistic optimization.

### 2. Strategic Optimization

### Framework and first-order condition
- With strategic optimization, AEs anticipate that their emissions will impact EMDEs’ subsequent emitting behavior.
- (A9)     
  퐶_AAA = 푔(푒_AAA + 푒_AEEE_A(푒_AAA))
- First-order condition:
  - (A10)     
    푓′(푒̅_AAA_strat) = 푔′(푒̅_AAA_strat + 푒̅_AEEE_A_strat) × [1 + (∂푒̅_AEEE_A_strat / ∂푒̅_AAA_strat)]
  - where str denotes strategic optimization.

### Comparative implications and ambiguity
- Strategic AEs compare the development benefit of emitting additional GHGs with the cost of additional overall GHG emissions that their expansion induces, not just the immediate cost of their own emissions.
- From (A3):
  - The increase in overall emissions will be less than the increase in AE emissions (i.e., induced emissions in EMDEs are smaller than the AE emission increase).
  - However, the induced increase occurs at a higher concentration of GHGs in the atmosphere and therefore is more costly per unit of emitted GHGs.
- Net effect:
  - Whether strategic behavior implies more or less emissions than individual optimization is ambiguous and depends on the model’s parameterization.
  - For the base specification used in the main text, the differences between individual and strategic optimization are minimal.

*Source: wpiea2021280-print-pdf - 1.   Altruistic Optimization*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021280-print-pdf.pdf_
