The Paris Agreement's goal of limiting global warming necessitates a rapid transition to low-carbon technologies. Many crucial technologies for this transition are not yet fully market-ready, highlighting the critical role of Research and Development (R&D) investment. Current climate neutrality assessments often neglect the contribution of research-driven innovation. This paper bridges this gap by integrating two established integrated assessment models to analyze optimal R&D investment strategies for various low-carbon technologies, including batteries for electric vehicles (EDVs), advanced biofuels, solar, wind power, and carbon capture and storage (CCS), along with energy efficiency improvements. The study also explores a financing mechanism using carbon revenues to fund this R&D, considering its macroeconomic and employment impacts. The importance of this research stems from the need for informed decision-making on public R&D funding to accelerate the energy transition and achieve climate stabilization targets at the lowest possible cost.

Existing literature strongly supports the positive impact of R&D investments on lowering energy technology costs. Studies consistently demonstrate that public R&D funding, especially "patient capital", is crucial for overcoming the uncertainties inherent in non-incremental energy innovation and fostering learning-by-research dynamics. Successful examples include the cost reductions achieved in wind, solar, LEDs, and EV batteries through state-funded innovation and supportive policy frameworks. However, choices in public R&D funding across different technologies at varying development stages require careful consideration, recognizing the complex interactions between various low-carbon options and regional specificities. The optimal timing of R&D investments is also crucial for maximizing effectiveness in limiting global warming. Most integrated modeling assessments of climate stabilization, however, overlook the financing mechanisms for R&D, a critical aspect addressed in this paper.

This study employs a soft-linked approach combining two integrated assessment models: WITCH and GEM-E3. WITCH, a dynamic model, identifies optimal R&D investments across five key decarbonization technologies and energy efficiency measures, considering technology substitution and complementarity. It optimizes R&D investments intertemporally, accounting for the time lag between investment and cost reductions. GEM-E3, a computable general equilibrium (CGE) model, evaluates the macroeconomic repercussions, including employment effects and competitiveness, of different R&D investment choices, considering a detailed representation of the economy with 67 production sectors. The models incorporate two-factor learning curves, reflecting cost reductions from both R&D investments and increased deployment (learning-by-doing). Knowledge spillovers are also integrated to account for the influence of foreign knowledge on domestic innovation. Learning rates for each technology were drawn from existing literature. Several scenarios were analyzed, including a reference scenario (REF) with no climate stabilization target and policy scenarios targeting 2°C and 1.5°C warming by 2100, with and without optimal R&D investments. In the policy scenarios with optimized R&D, carbon revenues are used to finance additional R&D investments, with any surplus used to reduce payroll taxes. The models were linked by using WITCH's results (emissions pathways, R&D knowledge stocks, and R&D investments) as inputs for GEM-E3’s analysis of the macroeconomic implications of different R&D investment strategies.

The analysis reveals several key findings: 1. **Optimal Technology Portfolio:** Regardless of the stringency of the climate target, batteries for EDVs and energy efficiency consistently receive the largest share of R&D investments. The relative importance of other technologies varies depending on the climate target. Under stricter targets (1.5°C), advanced biofuels receive a larger share, while under less stringent targets (2°C), solar and wind investments are more significant. CCS investments remain relatively low compared to others, especially in the less stringent scenario. 2. **Timing of R&D Investment:** For mature technologies like solar and wind, early investment timing is more critical than overall investment levels. Stringent climate policies lead to earlier investment in these technologies, maximizing their cost-reducing learning effects. For less mature technologies such as CCS, achieving stringent decarbonization requires substantial increases in R&D investment. 3. **Regional Investment Heterogeneity:** R&D investment varies significantly across regions, with the USA, EU, Japan, and South Korea being the largest contributors. Achieving the 1.5°C target requires substantially higher R&D investment in most regions compared to the reference scenario, especially in China, Japan, South Korea, Southeast Asia, Latin America, and sub-Saharan Africa. 4. **Financing R&D with Carbon Revenues:** The study demonstrates the feasibility of financing additional R&D investments using carbon revenues. Global carbon revenues in 2050 under the 2°C (1.5°C) target reach 2.5% (4.1%) of GDP. For most regions, additional R&D funding represents a small share of carbon revenues, with the remainder available to reduce payroll taxes. This creates a "double dividend," lowering mitigation costs and stimulating employment through reduced distortionary taxation. 5. **Macroeconomic Effects:** Optimal R&D strategies yield positive macroeconomic impacts, particularly under the 1.5°C scenario. Globally, GDP increases, driven by increased competitiveness in clean technology production and lower technology costs. This results in higher consumption and economic activity. Positive employment effects are also observed globally, despite the use of carbon revenues for R&D funding. 6. **Mitigation Cost Reduction:** Optimal R&D strategies significantly reduce the overall cost of achieving climate targets by lowering technology costs and reducing the required carbon tax. The cost reductions are more substantial for the more stringent 1.5°C scenario. For the 1.5°C scenario, each dollar invested in R&D resulted in an 8.01 dollar increase in GDP and 96 additional jobs in 2050.

The findings demonstrate the crucial role of R&D investment in achieving the Paris Agreement targets. The optimal R&D strategies identified in this study provide valuable insights for policymakers, indicating the need for a diversified portfolio of technologies, early investments in mature technologies, and substantial increases in R&D for less mature options, particularly CCS. The economic analysis underscores that this increased investment, funded through carbon revenues, not only reduces mitigation costs but also leads to positive macroeconomic and employment outcomes. The findings highlight that carbon pricing mechanisms can effectively finance R&D while generating wider economic benefits. The model results suggest the efficacy of carbon pricing in creating an incentive for both emissions reductions and clean technology development, particularly in promoting the 'polluter pays' principle while generating additional economic and social benefits.

This research provides a robust framework for optimizing R&D investment strategies to achieve ambitious climate targets. The combined use of WITCH and GEM-E3 models offers valuable insights into optimal technology portfolios, investment timing, and the feasibility of financing R&D through carbon revenues. The results demonstrate the significant economic and employment benefits of strategic R&D investment, highlighting the importance of integrating innovation into climate policy design. Future research could explore alternative financing mechanisms, further refine technology modeling, and investigate the role of international cooperation in fostering global R&D efforts.

The study's limitations include the specific set of technologies considered, which may not encompass all potentially relevant options. The reliance on existing literature for learning rates introduces a degree of uncertainty. Furthermore, the modeling framework assumes perfect foresight, which may not fully capture the complexities of real-world technological development. The study focuses on a specific carbon revenue recycling mechanism and does not explore other potential financing options. The model also does not explicitly address technological uncertainty beyond the calibrated learning rates.