Drivers of the US CO2 emissions 1997–2013
Feng, Davis, t. al., July 21, 2015 (Nature Communications)
Fossil fuel CO2 emissions in the United States decreased by B11% between 2007 and 2013, from 6,023 to 5,377 Mt. This decline has been widely attributed to a shift from the use of coal to natural gas in US electricity production. However, the factors driving the decline have not been quantitatively evaluated; the role of natural gas in the decline therefore remains speculative. Here we analyse the factors affecting US emissions from 1997 to 2013. Before 2007, rising emissions were primarily driven by economic growth. After 2007, decreasing emissions were largely a result of economic recession with changes in fuel mix (for example, substitution of natural gas for coal) playing a comparatively minor role. Energy–climate policies may, therefore, be necessary to lock-in the recent emissions reductions and drive further decarbonization of the energy system as the US economy recovers and grows.
The CO2 emissions from the burning of fossil fuels are the primary cause of anthropogenic climate change1, and the United States emits more CO2 each year than any other country except China. In the decade before 2007, US CO2 emissions grew by an average 0.7% per year. However, beginning in 2007, US emissions decreased, reaching a minimum of 5,284 Mt CO2 in 2012—12% lower than 2007 levels and 5% lower than 1997 levels2. This recent decline is good news and is consistent with the Obama administration’s stated goal of reducing CO2 emissions by 17% in 2020 and 83% in 2050 relative to 2005 levels3. Assuming no change in emissions outside the power sector, the new rules proposed by the US Environmental Protection Agency in June 2014 to limit CO2 emissions from power plants will require US emissions to decrease to 4,200 Mt CO2 in 2030—a further 20% reduction from 2013 levels4.
Coinciding with the post-2007 decline in emissions, innovations in hydraulic fracturing technology have dramatically increased domestic supplies of gas5,6. Commentators in the scientific community and media have linked the two trends, celebrating the climate benefits of the gas boom7–9. Recently, the Third National Climate Assessment of the United States Global Change Research Program also adopted this conclusion, stating that the decrease in US CO2 emissions was ‘ylargely due to a shift from coal to less CO2-intensive natural gas for electricity production’10. Yet, despite potentially significant implications for US climate and energy policy, there has been no quantitative analysis of whether the gas boom and changes in the fuel mix of the power sector are indeed driving the decrease in US CO2 emissions.
Here, we use input–output structural decomposition analysis (SDA) to assess sources of change in US CO2 emissions over a decade of mostly increasing emissions, 1997–2007, and then over the period of mostly decreasing emissions, 2007–2013. Our analysis quantifies the contribution of six different factors to changes in US emissions. These factors are: population growth; changes in consumption volume caused exclusively by changes in per capita consumption of goods and services; shifts in consumption patterns or the types of goods and services being consumed; adjustments in production structure or the mix of inputs (for example, labour, domestic and imported materials) required to produce US goods and services; changes in fuel mix as reflected by the CO2 emitted per unit of energy used; and changes in energy intensity or the energy used per inflation-adjusted unit of economic output. The SDA in this research is based on the additive decomposition of the changes in emission determined by six multiplicative factors acting as accelerators or retardants of the emission dynamics. Each term in the decomposition is a product of the change in one explicative factor and the level values of the other five factors, and thus represents the contribution of one explicative factor to the total change in emission. For example, in the term where population is the explicative factor, the values of consumption volume, production structure, consumption patterns, energy intensity and fuel mix are held unchanged and only population varies. In this way, the SDA method allows us to quantify the contribution of each of the assessed factors to the trend in emissions. Details of our methodology and data sources are in the Methods section (including Supplementary Methods). We find that before 2007, rising emissions were driven by economic growth: 71% of the increase between 1997 and 2007 was due to increases in US consumption of goods and services, with the remainder of the increase due to population growth. Concurrent with the global economic recession, 83% of the decrease during 2007–2009 was due to decreased consumption and changes in the production structure of the US economy, with just 17% related to changes in the fuel mix. During the economic recovery, 2009–2013, the decrease in US emissions has been small (o1%), with nearly equal contributions from changes in the fuel mix, decreases in energy use per unit of GDP, changes in US production structure, and changes in consumption patterns. We conclude that substitution of gas for coal has had a relatively minor role in the emissions reduction of US CO2 emissions since 2007.
Growing emissions from 1997 to 2007. Between 1997 and 2007, US emissions increased by 7.3% (Fig. 1, black curve). Our analysis shows that the main factor behind this increase was an increase in consumption volume caused by growth in per capita consumption of goods and services in the United States. Indeed, increases in such consumption volume correspond to a contribution of a 21.8% increase in emissions over this decade (Fig. 1, red curve). The next most important factor influencing CO2 emissions over the same period was population growth. Immigration and natural growth have resulted in steady population growth at a rate of B1% per year since 1997. These population gains contributed to an 8.9% increase in emissions between 1997 and 2007 (Fig. 1, yellow curve).
However, other factors slowed the growth of emissions between 1997 and 2007: decreases in the energy intensity of GDP; changes in the consumption patterns of US consumers; shifts in production structure; and decreases in the use of coal as an energy source. For instance, over this period, the energy used per dollar of economic output decreased by 17% (Fig. 2a, black curve), the share of consumer spending on manufactured goods decreased by B4% (Fig. 2b), the share of imported inputs to the US industry sectors increased (for example, imports to petroleum and coal products sector increased by 6.7%, and imports to the chemical products, primary metals and textile sectors increased by 2.7%, 2.5% and 2.1%, respectively)11, and the share of US electricity generated from coal decreased by B5% while the share generated from natural gas increased by 8% (Fig. 2c). All of these trends exerted a downward influence on emissions. Between 1997 and 2007, changes in energy intensity, consumption patterns, production structure and fuel mix contributed to retarding emissions of 7.4, 6.9, 4.9 and 3.6%, respectively (Fig. 1, purple, green, blue and orange curves, respectively).
Declining emissions from 2007 to 2013. US CO2 emissions stopped growing in 2007, and decreased by B11% between 2007 and 2013 (Fig. 1, black curve). Looking at this time period in aggregate, the only factor which acted to increase emissions over the period was continued and steady population growth ( þ 3.7%) (Fig. 1, yellow curve). However, the upward influence of population growth was overwhelmed by the downward influence of changes in production structure ( 6.1%), fuel mix ( 4.4%), consumption volumes triggered by per capita consumption ( 3.9%), energy intensity of GDP ( 0.5%) and changing consumption patterns ( 0.4%; Fig. 1, blue, orange, red, purple and green curves, respectively).
Although all of the analysed factors except population contributed to the decrease in emissions during 2007–2013, different factors dominated over shorter periods. Figure 3 subdivides 2007-2013 into 2-year periods, showing that emissions fell by 9.9% from 2007 to 2009, increased by 1.3% between 2009 to 2011 and decreased again by 2.1% between 2011 and 2013.
More than half (53%) of the initial and most substantial decrease in emissions, between 2007 and 2009, was due to a sharp drop in the volume of consumed goods as a result of reduction in per capita consumption during the global economic recession (Fig. 3, red bar). In particular, Fig. 4 shows that sharp decreases in the volume of capital expenditures and exported goods between 2007 and 2009 drove down associated emissions by 25% and 18%, respectively. Changes in the production structure of the US economy (that is, the volume and type of intermediate goods demanded) and the fuel mix of the energy sector contributed 30% and 17% of the initial (2007–2009) decrease in emissions, respectively, while increases in the energy intensity of the US economy and changing consumption patterns exerted modest upward influences on emissions during the same period.
As the US economy had slowly recovered from the global economic recession, between 2009 and 2013, the average annual change in US emissions was small: a 0.2% decrease. Economic recovery is reflected by the upward influence of the volume of goods consumed on emissions during both 2009–2011 and 2011–2013. Between 2009 and 2011, rising consumption volume, population growth, and increasing energy intensity urged emissions up by a combined 4.0% (2.2%, 1.5% and 0.3%, respectively), which was only partly offset by the changes in consumption patterns ( 1.1%), production structure ( 1.0%) and fuel mix ( 0.6%), resulting in an actual increase in emissions of 1.3% (Fig. 3). However, between 2011 and 2013, the upward influence of consumption volume and population on emissions was less ( þ 1.2% and þ 1.2%, respectively) and the energy intensity of the economy decreased ( 2.1%). When combined with changes in the fuel mix of the energy sector ( 1.2%) and shifting consumption patterns ( 0.2%), the net effect was a 2.1% decrease in emissions during 2011–2013 (Fig. 3).
Increases in the supply of natural gas affect two of the factors in our analysis: the fuel mix of the energy sector and, to a lesser extent, the energy intensity of the US economy. By decreasing gas prices, abundant gas encourages a shift in the fuel mix from more carbon-intensive coal to gas. In turn, a shift to gas may contribute to decreased energy intensity because gas-fired power plants are on average 20% more efficient at converting fuel energy to electricity than coal plants12.
The boom of natural gas from breakthroughs in hydraulic fracturing of shale deposits had only just begun to affect US gas supplies in 2009 (ref. 5). Thus, the decrease in emissions from changes in the fuel mix of the energy sector prior 2009 reflects an independent and longer-term trend of the declining use of coal in the US energy sector (see, for example, Fig. 2c). However, as seen in Fig. 3, changes in the US fuel mix from 2007 to 2009 alone would not have caused a decrease in US emissions.
Although the decreases in emissions since 2009 have been relatively small, the influence of shale gas is visible. For example, about half of the 2.1% decrease in emissions during 2011–2013 is related to changes in the fuel mix of the energy sector ( 1.2%, orange bar in Fig. 3). Yet the decrease in the energy intensity of the US economy was nearly twice as strong an influence on emissions over the same period (purple bar in Fig. 3).
Although a drop in the energy intensity (exajoule per dollar output) of the energy sector in 2013 accounts for roughly a third of the observed decrease in US energy intensity in 2011–2013, the remaining two-thirds relate to changes in energy used by the transport and service sectors (Fig. 2a). Three unrelated trends underlie the decreasing energy intensity of these sectors. First, high gasoline prices during 2011–2013 (the average price of gasoline had remained above $3.40 per gallon during this period, in contrast to the average price of $2.50 per gallon in 2005) have contributed to both a reduction in per capita miles driven (Supplementary Fig. 1a) and an increase in average fuel efficiency of vehicles (Supplementary Fig. 1b), and thus a 33% decrease in US gasoline consumption during 2011–2013. Second, a mild winter in 2012 meant less energy was used for heating and thus reduced energy intensity of the service sector (households also used less energy for home heating, which accounts for part of the drop in consumption volume)13 (Supplementary Fig. 2). Last, there is evidence that manufacturing in the United States became more energy efficient: energy use by manufacturing was nearly constant 2011–2013 despite average annual growth in GDP of 2.3% per year over the same period.
Shifts in the production structure of the US economy between 2007 and 2013 have consistently exerted a downward influence on US emissions, as the volume and type of intermediate goods used by various industry sectors has evolved and become more efficient (blue bars in Fig. 3). Yet this structural shift also reflects the progressive offshoring of emissions-intensive industries to China and other developing countries over the analysed period14. For instance, between 2009 and 2011, when changes in domestic production structure exerted a downward influence on US CO2 emissions ( 1%, blue bar in Fig. 3), we calculated that the net import of emissions embodied in US trade increased by 32% (Supplementary Fig. 3). Trade data for the 2011–2013 period is not yet available.
Between 2009 and 2013, the share of US consumption of manufactured goods increased relative to services (Fig. 2b), but the net effect of changes in consumption patterns was to decrease emissions (by 1.1% between 2009 and 2011 and by 0.2% between 2011 and 2013; green bars in Fig. 3). This result reveals that changes in the types of goods being consumed over time can have a significant impact on emissions15,16, and that it is not as simple as the balance of manufactured goods and services…
Decomposing Final Demand. Because changes in the volume of goods and services consumed were the single most important influence on US emissions between 1997 and 2013, we also analysed four separate components of final demand to assess the trends in emissions related to each category as well as the important influences on emissions in each case. Figure 4 shows the emissions associated with different final demand (consumption) components: household consumption (Fig. 4a), governmental expenditure (Fig. 4b), capital formation (Fig. 4c) and exports (Fig. 4d).
Between 2007 and 2013, emissions associated with household consumption decreased by 11.0%, which was almost entirely driven by changes in fuel mix and production structure, especially between 2009 and 2013, since consumption volume was constant (Fig. 4a). Emissions associated with government expenditures in the same time period decreased by 4.8%, and it was largely driven by changes in energy intensity and production structure (Fig. 4b). In contrast, emissions related to capital formation decreased by 24.4% between 2007 and 2013, primarily due to a huge decline in the volume of capital investment (Fig. 4c, red curve). However, changes in emissions related to exports between 2007 and 2013 were almost entirely the result of changes in the volume of exports, with the other factors cancelling each other out (Fig. 4d).