Saving less in China facilitates global CO2 mitigation

China’s domestic saving rate has been among the highest in the world (about 47% of GDP in 2017). High saving rates in China have historically moved in step with high investment rates, stimulating capital accumulation and demand for energy- and material-intensive capital goods (infrastructure, equipment, machinery), thereby raising CO2 emissions domestically and globally. Given China’s commitments under the Paris Agreement (peaking CO2 emissions around 2030) and the diminishing returns to investment-heavy growth, China is seeking to shift from investment-driven to consumption-driven growth. Policy initiatives and demographic trends (population ageing) are expected to reduce saving rates. This study asks: to what extent can changes in China’s saving rates contribute to global CO2 mitigation, and how do impacts vary across regions within China? The paper develops theoretical and empirical approaches to quantify the partial effects of saving rate changes on global industrial CO2 emissions and assesses region-specific potentials under different scenarios.

Existing research has extensively examined China’s role in global CO2 emissions, particularly the effects of changes in final demand structure, consumption patterns, and investment composition on emissions at national and regional scales. Studies have analyzed drivers of China’s emissions, consumption-based accounting for megacities, and unequal household carbon footprints. However, prior work has not investigated how changes in China’s saving rates per se affect global CO2 emissions. Moreover, saving rates vary markedly across Chinese regions (e.g., 37% in Shanghai versus 70% in Inner Mongolia in 2012), implying heterogeneous mitigation potentials. This study fills these gaps by linking saving rate changes to global industrial CO2 emissions, accounting for regional diversity.

The study combines: (1) a theoretical dynamic model and (2) an environmentally extended multi-regional input–output (EE-MRIO) framework with a new structural decomposition analysis (SDA), plus counterfactual scenarios.

1. Theoretical model: A stylized two-sector model (capital goods and consumption goods) with two representative firms and one representative household derives how changes in saving rates (s) affect steady-state outputs and CO2 emissions. Emissions are proportional to outputs via sector-specific emission intensities. The analysis shows that if the cumulative (direct+indirect) CO2 emission intensity of capital goods production is sufficiently larger than that of consumption goods, lowering s reduces total CO2 emissions.
2. EE-MRIO data and construction: The global MRIO table integrates GTAP MRIO data (57 sectors for 139 non-China economies) with a Chinese inter-provincial MRIO (30 sectors for 30 provinces). Sectoral CO2 emissions are attached as satellite accounts. The core balance is x=(I−A)−1 f, with emissions from production eP= q̂ (I−A)−1 f. Household direct CO2 emissions in China are estimated separately, assuming constant per-unit emissions by energy consumption structure.
3. Structural decomposition analysis (SDA): Changes in global industrial CO2 emissions between 2007 and 2012 are decomposed into contributions from (i) sectoral CO2 emission intensity, (ii) production input structure (Leontief inverse), (iii) China’s disposable income, (iv) China’s saving rate (via proportions of final consumption vs capital formation), (v) structures of final consumption and capital formation, and (vi) other final-demand factors. The saving rate effect is computed by summing contributions of changes in the proportions of final consumption and capital formation. Saving rate is defined as (GDP − consumption)/GDP; changes in saving and investment rates are treated as two sides of the same coin.
4. Counterfactual scenarios: Holding global supply chains (A), sectoral emission intensities (q), and income levels constant, the study alters saving rates to evaluate partial effects on global CO2:

• Scenario 1: Decrease saving rates in all Chinese regions by 15 percentage points (based on forecast decline from 47% to 32% during 2019–2035). Emission intensity fixed at 2012 values.
• Scenario 2: Decrease the saving rate of one Chinese region at a time by 15 percentage points (others unchanged). Emission intensity fixed at 2012 values.
• Scenario 3: Set each region’s saving rate to 37% (Shanghai’s 2012 level), one region at a time.
• Scenario 4: Scenario 1 plus “consummate greener consumption,” whereby, for each sector, the commodities consumed in China are hypothetically sourced from the nation/region with the lowest cumulative CO2 emission intensity (direct+indirect). An anomaly is excluded by replacing an abnormally low value for petroleum and coal products in Botswana with the second-lowest intensity.

5. Additional analyses: Global CO2 induced per unit basket of China’s final consumption vs capital formation is estimated by allocating one unit across sectoral shares and applying (I−A)−1. Sensitivity analysis calculates elasticities of results with respect to parameters (q, A/T, disposable incomes, consumption and investment structures, saving rate changes). Uncertainty ranges are derived from plausible bounds on China’s future saving rate trajectories. The study focuses on static comparative statics and does not model long-run economic growth or dynamic technology change.

• Historical SDA (2007–2012): China’s national saving rate rose from 44% to 53%. The increase in saving rates of Chinese regions raised global industrial CO2 emissions by 188.6–189 Mt, about 0.7% of global industrial CO2 in 2012, ceteris paribus. Effects were sectorally asymmetric: increments concentrated in capital-goods-related sectors (e.g., metallurgy +130.5 Mt, nonmetal products +33.7 Mt, construction +9.9 Mt within China; electricity +11.2 Mt, ferrous metals +2.3 Mt, other minerals +2.1 Mt abroad), while decreases occurred in consumption-oriented sectors (e.g., agriculture −12.8 Mt; food processing and tobaccos −7.6 Mt; other services −7.4 Mt in China; various agricultural products abroad). Country impacts from China’s saving rise included increases in Japan (+2.7 Mt), South Africa (+1.9 Mt), Australia (+1.8 Mt), South Korea (+1.7 Mt), India (+1.5 Mt), USA (+1.3 Mt), and small decreases in Argentina (−0.2 Mt), New Zealand (−0.02 Mt), Uruguay (−0.01 Mt). Within China, CO2 increased notably in Hebei (+26.3 Mt), Shandong (+23.4 Mt), Henan (+19.5 Mt), Liaoning (+14.7 Mt), Jiangsu (+14.7 Mt); decreases occurred in Heilongjiang (−5.5 Mt), Xinjiang (−1.9 Mt), Fujian (−0.7 Mt), Guizhou (−0.3 Mt), Guangdong (−0.3 Mt).
• Mechanism quantification: A unit of final consumption in China induces about 1.0 t of global CO2; a unit of capital formation induces about 1.5 t. Hence, shifting from investment to consumption reduces global CO2.
• Scenario 1 (−15 percentage points in all regions): Global industrial CO2 from production systems decreases by 323 Mt (−1.2%), while Chinese household direct emissions increase by 137 Mt due to higher consumption; net global reduction is 186 Mt (−0.7%). National reductions include Japan −4.1 Mt (−0.4%), South Africa −3.9 Mt (−1.2%), South Korea −2.9 Mt (−0.6%), USA −1.6 Mt (−0.04%), Russia −1.5 Mt (−0.1%), Germany −1.3 Mt (−0.2%). Increases occur in Argentina (+0.27 Mt), New Zealand (+0.14 Mt), Qatar (+0.05 Mt). Largest percentage reductions abroad: Zambia −2.9%, Chile −1.4%, South Africa −1.2%; largest percentage increases: Uruguay +0.6%, New Zealand +0.5%, Argentina +0.2%. Within China, production-system CO2 falls most in Hebei (−34.2 Mt), Jiangsu (−27.0 Mt), Liaoning (−24.1 Mt), Shandong (−15.4 Mt). Reduction rates are highest in Ningxia (−9.5%), Chongqing (−7.1%), Jiangxi (−7.0%). Increases occur in Heilongjiang (+21.4 Mt; +8.9%), Inner Mongolia (+9.6 Mt; +1.7%), Guangdong (+7.2 Mt; +1.6%), reflecting greater consumption of foods and textiles.
• Scenario 2 (−15 percentage points one region at a time): Most provinces’ saving rate reductions lower global CO2. Shandong’s reduction yields the largest global CO2 decrease (−29 Mt). Inner Mongolia’s reduction increases global CO2 by +25 Mt due to its consumption mix (notably electricity) having higher cumulative emission intensity than its capital formation mix.
• Scenario 3 (set each region to 37% saving rate): If all provinces’ saving rates fell to 37%, global CO2 changes total −153 Mt (−0.6% of global industrial CO2), composed of −321 Mt from production systems and +168 Mt from Chinese households. Largest provincial potentials: Shandong −35 Mt, Liaoning −18 Mt, Shanxi −17 Mt. Inner Mongolia’s reduction would increase global CO2 by +55 Mt.
• Scenario 4 (−15 percentage points plus “consummate greener consumption”): Compared with Scenario 1, global CO2 emissions are 3593 Mt lower, implying a potential 14% reduction in global CO2 via greening China’s consumption sourcing alongside lower saving. Within China, major opportunities lie in reducing cumulative emission intensities in other services (notably in Shandong, Jiangsu, Hebei, Guangdong) and electricity (notably in Henan, Jiangsu, Guangdong, Heilongjiang). Among foreign suppliers of consumer goods to China, South Korea and the USA show large reduction potentials.
• Regional heterogeneity: Provinces with heavy industry and construction (e.g., Hebei, Liaoning, Shandong) benefit most (larger reductions) from reduced capital formation; resource-rich regions (e.g., Yunnan, Sichuan) also see declines. Food- and textile-producing regions (e.g., Heilongjiang, Guangdong) can experience increases due to higher consumption demand.
• Sensitivity: Most parameter elasticities are small. The highest sensitivity is the CO2 intensity of Hebei’s Metallurgy sector (elasticity 0.14). Household CO2 changes scale linearly with saving rate changes; Inner Mongolia’s household-related sensitivity is largest (elasticity 0.09).

The study demonstrates a positive historical correlation between China’s saving rate increases and higher global industrial CO2 emissions, and conversely, that reducing saving rates (shifting from investment to consumption) can contribute to global mitigation. Because capital formation in China has higher cumulative CO2 intensity than final consumption, reallocating expenditure lowers emissions. The findings imply that consumption-promotion policies—though not designed for environmental aims—yield co-benefits for climate mitigation. However, to maximize benefits, greener consumption is essential: shifting consumption toward lower cumulative emission-intensity suppliers and improving domestic sectoral efficiencies (especially services and electricity) amplify reductions. Results highlight substantial regional disparities within China, motivating region-specific policies: for instance, targeting food processing in Shandong versus electricity efficiency in Inner Mongolia. International cooperation, technology transfer, and measures to reduce economic frictions can help China adjust production structures to meet increased consumption demand while minimizing emissions.

This paper links China’s macroeconomic growth pattern—specifically saving and investment behavior—to global CO2 mitigation. It develops a theoretical mechanism and an EE-MRIO-based SDA and counterfactual framework to quantify partial effects of saving rate changes. Empirically, higher saving rates in 2007–2012 increased global industrial CO2 by about 189 Mt; a 15 percentage point reduction in saving rates would lower global CO2 by about 186 Mt (0.7%), with pronounced regional heterogeneity. Greener consumption could deliver an additional 14% reduction in global CO2 relative to a baseline with unchanged sourcing. Policy implications include: promoting low-carbon consumption (e.g., product carbon labeling), improving efficiency in key sectors (services, electricity, construction, machinery), and removing frictions so production systems adjust to consumption shifts. Region-specific strategies are crucial. Future research should incorporate dynamic technological progress induced by investment, improve MRIO uncertainty quantification and inter-regional trade data, and explore interactions with long-run economic growth.

• Uncertainty in future saving rate trajectories drives result ranges: estimated global industrial CO2 reductions of 292.4–513.3 Mt and Chinese household CO2 increases of 124.3–218.2 Mt, yielding net total reductions of 168.1–295.1 Mt.
• MRIO data uncertainty: GTAP-based MRIO lacks published standard deviations; many MRIO databases do not provide full uncertainty of raw statistics, limiting quantitative uncertainty analysis.
• Model scope: Static comparative statics; long-run economic growth effects are not modeled. Dynamic consequences of investment on technological progress and energy efficiency are excluded.
• Chinese inter-regional trade in the MRIO relies on estimates based on inter-provincial railway transportation data; more accurate domestic trade surveys would improve precision.
• Sensitivity: While most elasticities are small, results are relatively sensitive to certain parameters (e.g., Hebei Metallurgy CO2 intensity, disposable incomes and saving rate changes in Shandong and Jiangsu).