Power-Sector CO2 Emissions Across the United States

Project Description

This analysis assesses the total volume of power-sector CO2 emissions between the years 1990 and 2020 for all U.S. States and the United States as a whole. It then correlates seven predictor variables with these CO2 emissions to identify potential influences on total power-sector CO2 emissions.

The Results: Key Takeaways & Policy Implications

For brevity, only results from the United States as a whole will be reviewed:

The total volume of CO2 Emissions increased by 30% from 1990 to 2007 and then declined by 39% by 2020.

An increase in average electricity cost was weakly associated with a decrease in total power-sector CO2 emissions. This is most likely due to a decreased demand for energy production with an increase in average electricity cost. This indicates that increasing energy costs could lower power-sector CO2 emissions, although this could have other negative implications for the economy as a whole.

An increase in net energy generation is weakly associated with an increase in power-sector CO2 emissions. However, coal use was strongly associated with power-sector CO2 emissions, indicating that reducing coal emissions may help reduce power-sector CO2 emissions. Fortunately, coal use has been declining in the United States since 2008. This coal use has been replaced with natural gas, which has a weak negative correlation with CO2 emissions, most likely because it produces less CO2 than coal.

21% of the total net energy generated in the United States in 2020 came from clean energy sources. An increase in clean energy sources as a whole was weakly associated with decreased power-sector CO2 emissions. However, Solar Thermal and Photovoltaic, Wind and Geothermal energy sources were strongly correlated with a decrease in CO2 emissions. Investing in these technologies could continue to lower power-sector CO2 emissions in the United States.

Population growth and an increase in population density were both weakly associated with decreased power-sector CO2 emissions, indicating a decoupling between population growth and population density and power-sector CO2 emissions.

Real GDP across industries is strongly associated with a decrease in power-sector emissions, indicating a decoupling between economic output and power-sector CO2 emissions. This was not the case for construction, for which an increase in utilization is associated with a weak increase in power-sector CO2 emissions. The link between increased construction and increased power-sector emissions warrants further investigation to determine root causes and potential solutions.

Median household income was weakly associated with a decrease in power-sector CO2 emissions, suggesting a decoupling between population wealth and power-sector CO2 emissions.

Total Federal renewable energy expenditures were weakly associated with a decrease in power-sector CO2 emissions, indicating these investments have some impact on reducing power-sector CO2 emissions. Continuing to provide federal investments in renewable energy could continue to decrease power-sector CO2 emissions. However, exploring why such investments are not strongly correlated with CO2 emissions reductions is important to improve investment efficacy.

Project Methodology

This analysis includes one outcome (dependent) variable and seven predictor variables:

(I) Outcome (Dependent) Variable:

Total Estimated Power Sector CO2 Emissions By U.S. State By Year (1990-2020)

This variable displays the total CO2 emissions (in metric tons) released by the electric power sector for each U.S. state and the United States between the years 1990 and 2020

(II) Seven Predictor Variables:

The seven predictor variables were organized into three categories:

(1) Energy Production and Use Predictors:

Average Electricity Price By Consumer Type (1990-2020)

The mean cost of electricity (in cents per kilowatt-hour) charged to consumers by consumer type (i.e. residential, commercial, industrial, other, total) for each U.S. State and the United States between the years 1990 and 2020
Net Energy Generation By Energy Type (1990 - 2020)

The total amount of energy produced (in megawatt-hours) minus the energy consumed in energy production processes by energy type (e.g. coal, natural gas, wind, solar PV) for each U.S. state and the United States between the years 1990 and 2020

(2) Socioeconomic Predictors:

3. Population Total (1990-2020)

The total number of individuals residing in each U.S. State and the United States according to the U.S. Census between the years 1990 and 2020. Data Quality Note: U.S. Census population totals were only available for every 10 years. Missing years were interpolated using a linear function between the data of two available years.

4. Population Density (1990-2020)

The number of people residing in each U.S. State and the United States per geographic land area (in miles squared).

5. Real Gross Domestic Product (GPD) By Industry Type (1997-2020)

The total monetary value of all final goods and services produced by each U.S. state and the United States adjusted for inflation (in 2012 chained dollars) by industry type (e.g. manufacturing, agriculture) between the years 1997 - 2020

6. Median Household Income By U.S. State By Year (1990-2020)

The middle number of the distribution of household income (in 2022 dollars) for each U.S. state and the United States between the years 1990 and 2020

(3) Policy Predictors:

7. Total Federal Renewable Energy Expenditures (1991 - 2020)

The total expenditures by the U.S. Federal government on renewable energy, including direct expenditures, tax expenditures and research & development expenditures (in 2022 dollars) from 1991 to 2020. This provides the total financial investment for renewable energy expansion across all federal policies (e.g. subsidies, grants, tax breaks). Data Quality Note: Total federal renewable energy expenditure data was not available for all analysis years. Missing years were interpolated using a linear function between the data of two available years. Also, federal expenditure data was in Fiscal Year instead of Calendar Year. Therefore, Fiscal Year was converted to Calendar year. Since available data started in 1990, there was not a full year's worth of data for the year 1990, resulting in this dataframe having data from 1991 - 2020.

Correlations were calculated using a Pearson Correlation, which results in a number from -1 to 1 with -1 being a strong negative correlation (i.e. the predictor variable is strongly associated with a decrease in power-sector CO2 emissions) and 1 being a strong positive correlation (i.e. the predictor variable is strongly associated with an increase in power-sector CO2 emissions).

Of note: predictor variables were selected based on a review of the literature and public data availability. This analysis could incorporate more predictor variables in the future to help further understand the various contributors to energy-related carbon dioxide (CO2) emissions. This analysis could also be updated to incorporate more recent years of data. Further, new policy predictors, such as the impact of utility-scale PV subsidies, could be incorporated into future renditions of this project.

Project R Code:

DOWNLOAD

Project Data Files:

DOWNLOAD

The Results: In Detail

For brevity, I will only describe the results from the United States as a whole instead of the results of each State. Please ensure "United States" is selected from the drop-down menu at the top of each data dashboard to ensure the analysis description matches the data you are viewing. Please use this analysis process to explore the State data as well.

OUTCOME (DEPENDENT) VARIABLE:

For the United States, the total volume of CO2 Emissions increased 30% from 7,814,795,768 metric tons in 1990 to a peak of 10,188,129,994 metric tons in 2007 and then declined by 39% to 6,214,343,272 metric tons in 2020.

Now we'll examine the strength and direction of the relationship between these CO2 emissions and seven predictor variables over time to examine how a change in each predictor variable may be associated with changes in power-sector CO2 emissions.

(1.) AVERAGE ELECTRICITY PRICE BY U.S. STATE BY CONSUMER TYPE | ENERGY PRODUCTION AND USE PREDICTOR :

For the United States, the Average Electricity Price across all consumer types rose 226% from 13.4 Cents per kilowatt hour in 1990 to a high of 43.71 Cents per kilowatt hour in 2020. Commercial, Industrial and Residential consumers all paid significantly more for electricity in 2020 compared to 1990. Residential consumers paid the highest cost in 2020 at 60.74 Cents per Kilowatt-Hour, and Industrial consumers paid the lowest cost in 2020 at 27.23 Cents per Kilowatt-Hour.

All consumer types experienced a weak negative correlation between average electricity cost and power-sector CO2 emissions, meaning an increase in average electricity cost was weakly associated with a decrease in total power-sector CO2 emissions. This suggests that raising the cost of electricity can help to lower power-sector CO2 emissions, most likely by decreasing electricity demand. However, this may come at a cost, such as decreased quality of life. Therefore, it may not be the best means of lowering power-sector CO2 emissions.

(2.) NET ENERGY GENERATION BY U.S. STATE BY ENERGY TYPE AND BY CLEAN ENERGY INDICATOR | ENERGY PRODUCTION AND USE PREDICTOR

For the United States, total net energy generation rose 32% from 6.08 Billion Megawatt-Hours in 1990 to 8.02 Billion Megawatt-Hours in 2020. Coal production was the largest source of net energy generation from 1990 to 2015, at which time natural gas overtook coal as the leading source of net energy generation. Nuclear also slightly surpassed coal production in 2020. Wind generation increased 1,798% from 35.62 Million Megawatt-Hours in 2005 to 675.88 Million Megawatt-Hours in 2020, although this only comprised 8% of total Megawatt-Hours produced in 2020. Solar Thermal and PV also experienced a 1,963% increase from 8.65 Million Megawatt-Hours in 2012 to 178.40 Million Megawatt-Hours in 2020, although this only comprised 2% of total Megawatt-Hours produced in 2020. All other renewable energy sources remained relatively constant between 1990 and 2020. Therefore, while there has been a definite increase in renewable energy generation in the past 20 years, there is a long way to go before renewable energy makes up the majority of net energy generation in the United States.

Overall, there is a weak positive correlation between net energy generation and power-sector CO2 emissions, reflecting that net energy generation plays a role in increasing power-sector CO2 emissions. Coal reflected a strong positive correlation, and petroleum and other gases reflected a weak positive correlation with power-sector CO2 emissions, showing that these energy sources are associated with an increase in power-sector CO2 emissions. Alternatively, Solar Thermal and PV, wind and geothermal showed a strong negative correlation with power-sector CO2 emissions, and other renewable energy sources showed a weak negative correlation with power-sector CO2 emissions. Therefore, transitioning from coal and other fossil fuels to renewable energy, particularly solar thermal and PV, wind and geothermal may have a positive effect on lowering CO2 emissions, and further weaken the positive correlation between net energy generation as a whole and power-sector CO2 emissions. Suprisingly, natural gas had a weak negative correlation with power-sector CO2 emissions, suggesting it actually lowers CO2 emissions overall. This could be because natural gas has been replacing coal, which has greater power-sector-related CO2 emissions than natural gas, although these findings warrant further investigation.

For the United States, total net energy generation from unclean energy sources such as fossil fuels rose 39% from 10.36 Billion Megawatt-Hours in 1990 to a peak of 14.40 Billion Megawatt-Hours in 2007 and then fell 10% to 12.98 Billion Megawatt hours in 2020. This trend generally followed total net energy generation, as seen in the total net energy generation dashboard above. Clean energy sources such as wind and solar thermal and PV rose 70% from 1.80 Billion Megawatt-Hours in 1990 to 3.06 Billion Megawatt-Hours in 2020. However, while progress has been made in increasing the total amount of energy derived from renewable sources in the United States since 1990, only 21% of the total net energy generated in the U.S. came from renewable sources in 2020. Given the climate crisis, there is a long way to go to increase the share of renewable energy in the United States by 2050, when the United States has pledged to be carbon-neutral.

Overall, there is a weak positive correlation between net energy generation from unclean energy sources and power-sector CO2 emissions, reflecting that net energy generation from unclean energy sources is associated with increased power-sector CO2 emissions. Alternatively, there is a weak negative correlation between net energy generation from clean energy sources and power-sector CO2 emissions, reflecting that net energy generation from clean energy sources is associated with decreased power-sector CO2 emissions. However, these correlations are both weak, warranting further investigation into why clean energy does not have a strong correlation with decreasing power-sector CO2 emissions.

(3.) TOTAL POPULATION (1990-2020) | SOCIOECONOMIC PREDICTOR:

For the United States, the total population rose 33% from 248.71 Million in 1990 to a peak of 331.45 Million in 2020. Therefore, while total power-sector CO2 emissions have been decreasing since 2007, this trend was not reflected in total population size, suggesting that population totals do not have a particularly strong impact on power-sector CO2 emissions as other predictor variables such as the type of net energy generation. This trend is reflected in the correlation between population totals and total power-sector CO2 emissions, which has a weak negative correlation.

(4.) POPULATION DENSITY (1990-2020) | SOCIOECONOMIC PREDICTOR:

For the United States, total population density rose 33% from 70.41 people per square mile in 1990 to a peak of 93.83 people per mile squared in 2020. As with the total U.S. population, population density continued to increase between 1990 and 2020, while total power-sector CO2 emissions began decreasing since 2007. This suggests population density does not have a particularly strong impact on power-sector CO2 emissions as other predictor variables such as the type of net energy generation for the United States as a whole. This trend is reflected in the correlation between population density and total power-sector CO2 emissions, which has a weak negative correlation.

(5.) REAL GROSS DOMESTIC PRODUCT (GDP) BY INDUSTRY TYPE (1997-2020) | SOCIOECONOMIC PREDICTOR:

For the United States, Real GDP rose 57% from 11.65 Trillion Dollars in 1990 to 18.33 Trillion Dollars in 2020. Finance, insurance, real estate and leasing comprised the highest proportion of Real GDP during this entire timeframe. The Military comprised the lowest proportion of Real GDP throughout this timeframe.

Overall, a strong negative correlation exists between Real GDP and power-sector CO2 emissions, reflecting that an increase in real GDP is associated with decreased power-sector CO2 emissions. However, specific industries correlate with increased CO2 emissions, namely construction and the military, which both have a weak positive correlation with power-sector CO2 emissions. The "Other Services" category, which is nondescriptive, had a strong positive correlation with power-sector CO2 emissions, meaning an increase in this category is strongly associated with an increase in power-sector CO2 emissions. Most other Real GDP categories had a strong negative correlation with power-sector CO2 emissions, suggesting that these industries are strongly associated with decreased power-sector CO2 emissions.

(6.) MEDIAN HOUSEHOLD INCOME BY U.S. STATE BY YEAR (1990-2020) | SOCIOECONOMIC PREDICTOR:

For the United States, median household income rose 25% from $61,500 per year in 1990 to $76,660 per year in 2020. Therefore, median household income increased between 1990 and 2020, while total power-sector CO2 emissions decreased since 2007. This suggests that median household income does not have a particularly strong impact on power-sector CO2 emissions as other predictor variables, such as the type of net energy generation. This trend is reflected in the correlation between median household income and total power-sector CO2 emissions, which has a weak negative correlation.

(7.) TOTAL FEDERAL RENEWABLE ENERGY EXPENDITURES (1991-2020) | POLICY PREDICTOR:

Total Federal renewable energy (RE) expenditures rose 869% from $1.6 Billion in 1990 to $15.5 Billion in 2020. Total federal RE expenditures reflected a weak negative correlation with power-sector CO2 emissions, meaning an increase in Total RE expenditures was weakly associated with a decrease in power-sector CO2 emissions.