Conditions for low-carbon green growth

no clear strategy for materialising green growth, i.e. vastly reducing greenhouse gas emissions without diminishing economic growth, has been outlined. Here, we describe the conditions needed for green growth under a wide range of carbon budgets. The results indicate that integration of multiple socioeconomic transformative measures would support green growth, including lowering energy demand, shifting to an environmentally friendly food system, technological progress on energy technologies and the stimulus of capital formation induced by green investment. No single measure is sufficient to offset mitigation costs fully, indicating that holistic societal transformation is needed, as the realisation of all measures depends on effective government policies as well as uncertain social and technological changes.


Introduction
Economic growth that coincides with consideration of environmental protection or conservation, known as green growth or low-carbon green growth, has long been discussed 1 . Green growth is not a simple and well-defined concept but, conventionally, encompasses cessation of environmental degradation, consideration of natural capital and general promotion of sustainability or sustainable development 2, 3, 4, 5 . Over the last decade and especially since the Paris Agreement (PA) 6 , green growth has been a central objective of national and international organisations addressing climate change 7, 8, 9 .
The PA defines an international long-term climate change mitigation goal of limiting the increase in global average temperature to well below 2°C above pre-industrial levels and encourages pursuing efforts to limit the temperature increase to 1.5°C above pre-industrial levels. Along with the PA, national-level climate policies have developed rapidly. Nationally Determined Contributions (NDCs) outline short-term greenhouse gas (GHG) emissions reduction goals and, recently, NDCs have changed rapidly via two main channels. One of these channels is related to the long-term strategies submitted to the United Nations Framework Convention on Climate Change (UNFCCC) in achieved 21,22 . While there are some indications (or hopes) that greening of the economy may stimulate the economy and lead to structural changes that, in turn, have positive economic impacts, the literature addressing this topic to date remains rather limited and unclear about the types of efforts or policies required.
Here, we show the conditions needed for green growth under a wide range of stringent carbon budgets spanning global mean temperature increases of 1.5 to 2.0°C relative to the pre-industrial level. The conditions are based upon climate change mitigation scenarios that assume carbon pricing, as well as additional societal changes. To capture the effects of a wide range of such changes, we considered five major social transformations, namely, lowering energy demand 23 in conjunction with enhancement of electrification 24 , technological progress in the energy supply system leading to renewable and carbon capture and storage (CCS) cost reduction 24 , shifting to environmentally friendly food consumption including low-meat diets and reduction of food waste 25, 26 , stimulus of capital formation induced by green investment 27 , and implementation of all of these measures. We designated these scenarios "Energy-Demand-Change (EDC)", "Energy-Supply-Change (ESC)", "Food-System-Transformation (FST)", "Green-Investment (GI)", and "Integrated-Social-Transformation (IST)" respectively. year in the literature 29 for a carbon budget of 1000 Gt CO 2 , which is considered a cumulative mitigation cost expressed as net present value (NPV), and our estimates fall within this range (see red circle in Figure 1a). These costs are associated with additional energy system costs related to decarbonising the energy system, non-CO 2 emissions abatement and economic structural changes.
The mitigation cost is inversely correlated with the carbon budget, which is consistent with previous reports 30 . The periodic mitigation cost over this century is illustrated in Figure 1c. Mitigation costs are relatively large in the first part of this century, while the absolute cost (not relative to GDP) increases continuously over time (see Supplementary Figure 1). This periodic tendency is apparent regardless of carbon budgets and, as the budget becomes tighter, the magnitude of the cost increases The costs of climate change mitigation can be moderated through societal measures, which are presented in Figure 1. Full implementation of all social transformation measures allows mitigation costs to reach almost zero or even become negative for most carbon budgets, indicating that the green growth condition is met (Figure 1a). The scenarios in which carbon budgets are larger than 700 Gt CO 2 have negative mitigation costs, meaning mitigation would be beneficial over inaction.
As the carbon budget tightens, the degree of the GDP recovery decreases. For the budget of 500 Gt CO 2 , 3.9% recovery occurs from the default case and the offset effects are smaller than under a budget of 1000 Gt CO 2 . Thus, a larger carbon budget may provide a better opportunity to abrogate completely the GDP loss associated with climate change mitigation. This finding leads to the interesting conclusion that stronger climate mitigation goals will make it more difficult to achieve green growth.
In some cases, the early part of this century exhibits GDP losses, but the cost approaches the Cost decreases for renewable energy production (e.g. solar and wind) are often considered the largest factor in green growth. Our results indicate that such changes may be part of the growth drivers, but their contribution is limited. More importantly, their effects in our scenario are more prominent in the short term than the long term. Investment effects are essentially driven by cumulative capital inputs, which would be largest in the second half of the century (Figure 1c). These changes result in increased activity levels, mainly in the industrial and service sectors, while productivity decreases slightly ( Figure 2h). This productivity decrease occurs because labour is fixed and only capital is added, which causes an imbalance in production compared with the default case.
The Energy-Supply-Change condition primarily induces cost reductions in electricity generation, resulting in a relatively large share of energy being renewable. Then, the average electricity price decreases, which increases electricity demand, leading to an increase in activity levels ( Figure 2bc). This energy price decrease is beneficial to all sectors and, therefore, productivity rises. In particular, indirect effects on the service sector are the main driver of GDP recovery ( Figure   2i). Energy-Supply-Change includes two main pathways for moderating mitigation costs, namely, cost decreases for renewable energy and CCS. We examined which factor, renewable energy or CCS, is the major player in GDP recovery by modelling sensitivity scenarios to isolate these factors. The results show that the renewable energy and CCS cost decreases account for recovery of 0.7% and 0.3% of GDP respectively, indicating that cost decreases related to renewable energy would have a stronger influence than CCS.
The Energy-Demand-Change scenario decreases the demand for fossil fuels ( Figure 2d) and enhances electrification, which reduces the volume of "other energy supply" (Figure 2j). Two factors facing the power sector may offset recovery, namely, electrification and energy savings (Figure 2e), but the results indicate decreases related to these processes. The magnitude of the predicted changes is small relative to other energy supply factors. This supply-side energy decrease causes capital and labour to shift to other industries, supporting GDP recovery. The contributions to GDP recovery varied among energy demand sectors (industry, transport, and service), but the original sectoral scale appears to determine the magnitude of GDP recovery, making the service sector effect prominent.
The Food-System-Transformation condition includes three pathways for lowering mitigation costs. First, reductions in livestock-based food demand and food waste (Figure 2fg) directly reduce the demand for food production, leading to low mitigation costs for non-CO 2 (CH 4 and N 2 O) emissions from the agricultural sector (Supplementary Figure 7). Second, decreases in meat demand lessen demand for pasture area, which expands the potential for afforestation. Third, a portion of the production factors, labour and capital used for production activities in the agricultural sector under the default scenario, could be transferred to more productive sectors, such as the manufacturing and service sectors, thereby increasing total economic productivity. Small agricultural activity decreases are apparent under this scenario, which are eventually offset by service sector increases (Figure 2k).
The total effect of Food-System-Transformation over this century is not as large as that of energy system transformation in terms of GDP loss recovery; however, the decreases in CH 4 and N 2 O emissions contribute to reduced total GHG emissions, causing small decreases in the global mean temperature increase at the end of this century (Supplementary Figure 8).
In the integrated scenario, these effects are generally additive, and the interaction effects are small (Figure 2l). A similar trend was apparent in 2050 and 2100, as well as under other carbon budgets (Supplementary Figure 9).  (panels a, b, c, d, e, f and g, respectively). Panels h, i, j, k, and l show decomposition analyses of GDP recovery by sector. The black circles indicate the total net impacts on GDP recovery by sector.

Regional implications
The implications of social transformative measures differ among regions (Figure 3a). The

Figure 3
Regional implications of social transformation. a) Regional cumulative GDP loss rates expressed as NPV. b) Regional GDP loss recovery relative to the default scenario by region under a 1000-Gt budget. c) Regional carbon and energy intensity (units, kgCO 2 /$ and MJ/$). d, f) Shares of value added by the energy, industrial, and agricultural sectors. Regional definitions are provided in Supplementary Note 1.

Sensitivity analysis
The discount rate has long been a controversial topic related to the economics of climate change, and our results are also sensitive to assumptions related to this factor. At the end of this century, a discount rate of 3% leads to zero or negative mitigation costs under the Integrated-Social-Transformation scenario, as discussed above (Figure 4c). A discount rate of 1% yields greater gains, whereas 5% shows a small positive mitigation cost (0.1 to 1.1%). In contrast, the results for 2030 and 2050 show consistently positive values from 1.9 to 2.9% and 1.0 to 2.4%, respectively, regardless of mitigation level (Figure 4ab). NPV results based on discount rates depend on the difference between periodic mitigation cost trajectories and exponential curves, which has two main implications for this analysis. First, in the long term, social transformation can carry almost zero or negative mitigation cost, thereby meeting the condition of green growth, even with high discount rates. Thus, within the context of inter-generational considerations, the mitigation cost can be either moderated or increased by those measures. Second, in the short term, attaining net zero or negative conditions will be difficult. Thus, a clear trade-off exists between inter-generational and short-term considerations.
In our main analysis, we assumed that stringent mitigation efforts would begin immediately in 2021 but, until 2030, current NDCs might pin the emissions reductions to certain levels 32, 33, 34 . We tested scenarios incorporating the current NDCs and confirmed that the overall results are similar to the main results, but small differences were observed (Figure 4de). NDCs postpone the emissions reduction to later periods and may decrease short-term mitigation costs, but do not affect the GDP recovery level or the qualitative conclusions discussed above. indicating that societal transformation from multiple angles is required.
We defined the green growth condition from the perspective of GDP growth. It is also useful to focus on household consumption rather than GDP, which includes capital formation and net trade volume, as household consumption might be more relevant to human welfare. Naturally, Green-Investment directly boosts production through capital formation, while consuming some income that otherwise would have been used for household consumption. Therefore, the green growth condition, as defined based on household consumption, was not met under the scenarios in this study (Supplementary Figure 10). This finding suggests that stronger measures than were included in our scenarios are needed to realise green growth defined by household consumption rather than economic growth.
The impacts of climate change are the elephant in the room in the context of green growth and, therefore, they have been intensively reported and addressed in several recent articles 17, 35, 36, 37 that consider some aspects of the green growth concept. The damage function of the economic loss or growth associated with the temperature changes reported in some studies may be equivalent to or even greater than the climate change mitigation costs, indicating that economic growth would not be harmed by emissions reductions if climate change impacts in the baseline scenarios are considered.
However, due to the nature of the delayed response of the earth system, short-term temperature changes would not differ greatly, even with steep emissions reductions. Therefore, the qualitative conclusions of this study would not differ for the short term. Moreover, incorporation of the impacts of temperature change on GDP would strengthen our argument, increasing the advantage of climate change mitigation actions.
Similarly, a co-benefit of air pollution reduction associated with the GHG emissions reduction has often been noted as an additional source of green growth 38,39,40 . Incorporation of this factor into the green growth accounting would have different implications from the impacts of climate change, as the reduction in air pollution associated with climate change mitigation primarily carries short-term benefits. Although the benefit of avoiding premature death is often associated with the Value of Statistical Life (VSL) and accounted as an economic benefit, the actual economic market impacts would be limited 41 .
One point that has been discussed in the literature but not addressed in this study is the inequality and employment conditions associated with growth 22 . Unfortunately, directly addressing these factors in our modelling framework would be difficult. Notably, unemployment is more relevant to short-term than long-term conditions. The inequality implications of climate change mitigation would depend on the carbon tax recycling scheme 42 . Moreover, green growth itself could be defined more broadly to account for natural capital 19 , but we could not do so in this assessment.
For example, ecosystem benefits such as biodiversity conservation should be considered but were beyond the scope of our study.
Assuming that green growth is achievable, as shown in this study, the next question is how to transform society. Obviously, technological progress and innovation must play critical roles. The government could promote these improvements by changing the existing tax system or other regulations, which would lead to changes such as increased research and development expenditures for greening the economy. Another possible mechanism involves leadership guiding the direction of society to promote technological innovation. This process would require not only specific environmental policies but also broader industrial policies that consider carbon neutrality. Food system transformation, again, may rely on technological improvements, such as the development of artificial meat. However, more importantly, the environmental and health consciousness of individuals would be critical to reducing meat consumption 43,44 . For green investment, the assumptions in our scenarios might be interpreted as unrealistic. However, serious concern for future generations could lead to prioritisation of future consumption and savings of current money, providing many opportunities to change investment behaviour via Environment, Social and Governance (ESG) policies. In that sense, behavioural changes in investment occur naturally with changes in environmental and inter-generational consciousness.
Our findings open many new avenues for further research. The central question of such research is how the societal transformation assumed in this study can be realised. This could be addressed through modelling that extends the current framework by incorporating more granularity in the sectoral and regional data, or by improving the realism of the energy and food demand models used to assess feasibility. These changes may require additional data collection, including microdata such as household or industrial surveys. Whether behavioural changes in saving and investment associated with environmental consciousness will occur, and the degree of such changes, remain open topics for discussion. These factors are related to the on-going discourse over short-term and long-term green growth. It might be a straightforward assumption that richer people place more priority on future generations, while such prioritisation would be very challenging for the poorest people. In that context, promoting solutions to poverty and development issues may indirectly contribute to the realisation of green growth, and is thus a possible application of carbon tax revenue 42 .
among global AEZs (agro-ecological zones). Non-energy-related emissions from sources other than land-use changes are assumed to be proportional to the level of each activity (such as output). CH 4 has a range of sources, led by rice production, livestock, fossil fuel mining, and waste management.

Scenarios
We employed a two-dimensional climate change mitigation scenario framework, as described above (Supplementary Table 1). The stringency of climate change mitigation is represented by carbon budgets ranging from 500 Gt CO 2 to 1400 Gt CO 2 at increments of 100 Gt CO 2 to determine the effects of mitigation level in relation to the Paris Agreement, which suggests limiting global mean temperature in 2100 to well below 2°C or 1.5°C. Climate actions are assumed to occur immediately, beginning in 2021, with uniform global carbon prices (Supplementary Figure 11). In the sensitivity analysis, we analysed scenarios meeting the NDC emissions targets by 2030 and then switched to global climate action with a uniform carbon price (Supplementary Figure 11). NDC pledges limit carbon budgets based on feasibility 59,60 , and here we implement a 1000-Gt CO 2 scenario for comparison with the default immediate action scenarios.
Scenarios were analysed that represent types of social transformation to explore the effects of social transformations on climate change mitigation cost. We tested four social transformations, namely Energy-Demand-Change, Energy-Supply-Change, Food-System-Transformation, and Green-Investment. Conventionally, these changes are not represented as responses to carbon pricing in integrated assessment models an d a r e , i n s t e a d , t r e a t e d a s i n dependent socioeconomic assumptions; however, we associated them with emissions reduction measures, which, in turn, had significant impacts on GHG emissions and the macroeconomy.  23 , the reduction in energy demand is not as large in this study, but may nonetheless have meaningful impacts on the macroeconomy.
The Energy-Supply-Change scenario explores the possibility that energy supply-side technological progress is accelerated, specifically in relation to low-carbon energy. Costs associated with renewable energy generation (e.g. PV and wind) and storage of variable renewable energy (e.g. batteries) decrease more sharply than for the default case (Figure 2bc). In the meantime, CCS-related technology improves similarly, and the cost assumption is half of that in the default case. Such rapid technological progress is uncertain and cannot be easily attained by design. However, general environmental awareness and governmental leadership toward a carbon-neutral society would motivate companies involved in the development of these technologies to improve performance, which would eventually lead to cost reduction. Numerically, here we adopted the SSP1 assumptions for supply-side energy parameters 55 . We illustrate the primary energy supply in each scenario under a budget of 1000 Gt CO 2 in Supplementary Figure 12.
Food-System-Transformation focuses on environmental (and health) awareness by the public in conjunction with actual implementation, rather than technological improvement. In our scenarios, we assumed that livestock-based food consumption is restrained and food waste is reduced (Figure 2fg).
For livestock-based food consumption, calorie consumption is cut in developed countries and increases moderately in developing countries. For food waste, consumption-side food waste generation is halved as each Sustainable Development Goal is met. Recently, some reports have indicated that a healthy diet could also provide benefits to the environment 44 , and the dietary shift in this scenario meets both of those goals.
Green-Investment is a scenario wherein more priority is placed on future generations, and consequently, some current consumption is shifted to investment. Numerically, incremental 1% capital formation is added to the default case, which is assumed to last throughout this century.
These behavioural changes in saving and investment would involve stimulating the on-going shift to environmentally responsible investment, with more focus on ESG factors and general awareness in the population.

Decomposition analysis of GDP loss recovery
We conducted decomposition analysis of GDP loss recovery using the formula below.