Blog Post < Previous | Next >
Wikipedia
The divergent economics of electric utilities
If the World was straightforward, you would expect that electric utility rates would be driven by their fuel cost to generate the electricity.
If electric utility A has a higher fuel cost than electric utility B, electric utility A should pass on higher electricity rates to its customers than electric utility B.
Within this article I am going to explore how the above assumption holds up in the real World.
The cost of different fuels
Comparing fuel costs is not straightforward. However, there is some convergence between different sources. So, I looked at several of them.
First, let’s look at a time series going back to 2009 to 2023 from Lazard shown below.
Second, see a chart from Wikipedia covering the time frame from 2010 to 2019 shown below.
Original
sources include IRENA 2020 for renewables, Lazard for the price of
electricity from nuclear and coal, IAEA for nuclear capacity, and
Global Energy Monitor for coal capacity.
Third, see a bar chart from Statista using data as of 2021 shown below.
The three different charts are convergent. The main takeaways include:
a) The renewables, Solar & Wind, have become the most cost-effective fuels;
b) Nuclear has become increasingly expensive and not competitive with renewables;
c) Coal is now much more expensive than the renewables, but is far cheaper than Nuclear;
d) Depending on the data source, natural gas seems pretty competitive, not far off the renewables.
Fuel costs vs. utility rates expectations
In view of the above, we would expect that:
a) Utilities generating a higher percentage of their electricity using renewables (Solar, Wind) would offer lower rates to their customers;
b) Utilities generating a higher percentage of their electricity using Nuclear would have to charge higher rates to cover their higher generating costs;
c) Utilities generating a higher percentage of their electricity using Coal would have to charge more than the ones relying more on renewables.
Of course, things get a bit more complex because utilities use a mix of fuels to generate their electricity.
To check if the above assumptions hold up in the real world, I looked at State level data (for all 50 States), and focused on their respective energy fuel mix vs. the electricity price charged to customers (for all sectors combined).
Data source:
For fuel mix, I used the data from the Nuclear Energy Institute (NEI) as of 2021 (most recent data available). Oddly enough, they had by far the best data for fuel mix at the State level.
For the electricity price charged to customers (for all sectors combined), I used Table 5.6.A as of June 2023 from the US Energy Information Administration (EIA).
Fuel mix vs. utility rates in the real world
Relationship between fuel mix and electricity price to customers
Studying this relationship with scatter plots and LOESS models
The six scatter plots below study this relationship. The scatter plots are grouped into three different categories:
- Renewables: Solar and Wind;
- Zero emission: Nuclear and Hydro;
- Fossil fuels: Coal and Natural Gas.
The scatter plots above use a LOESS model to fit a curve that outlines the underlying trend in the data. It uses a 90% confidence interval (not to be confused with a prediction interval) to convey the uncertainty associated with the model’s fitted curve data point estimates.
The scatter plots X-axis, from left to right, shows a rising % of a given fuel within the overall generating mix of State utilities.
The scatter plots Y-axis shows the corresponding electricity rate in cents per kWh charged to customers.
Each data point represents a State. And, it conveys the % of a given fuel within the overall fuel mix for a given State (X-axis). It also conveys the average electricity price charged to customers in cents per kWh for that State (Y-axis).
As an example, within all the scatter plots you can see an outlier where one specific State charges nearly 40 cents per kWh. That is Hawaii. Reading through all the scatter plots, you can observe that Hawaii’s fuel mix is roughly:
- Solar 5%
- Wind 10%
- Nuclear 0%
- Hydro 0%
- Coal 12%
- Natural gas 0%
Hawaii is an outlier every which way as it relies mostly on petroleum for its electricity generation. It is the only State to materially rely on petroleum. If you visualize and extract the above information for Hawaii, you understand the information conveyed by these scatter plots.
So, what do the scatter plots tell us?
They tell us a lot about the relationship between specific fuels and prices or rates charged to customers. This is not to be confused with a specific fuel cost to a utility.
a) Coal, Wind, and Hydro are associated with the lower rates passed on to the customers regardless of their respective costs. As they represent a rising and dominant % of the fuel mix, electricity rates to customers converge towards 10 cents per kWh;
b) Natural gas and Nuclear translate into higher prices or rates passed on to customers. As they represent a rising % of the fuel mix, electricity rates to customers converge towards 15 cents per kWh;
c) Solar is associated with by far the most expensive prices or rates. As it represents a rising % of the fuel mix, electricity rates to customers converge towards 20 cents per kWh.
To better visualize these price differentials, let’s group them within their respective mentioned price tiers.
When focusing on the most competitive tier converging towards 10 cents
per kWh, we observe that there is one fuel from each of the three
different categories: Renewables-Wind, Fossil fuels-Coal, and Zero
emissions-Hydro.
Within the above table let’s focus first on Coal. The top 10 States that have the greatest % of Coal within their electricity generation mix charge in average 10 cents per kWh. The bottom 10 States charge 18.4 cents per kWh.
The overall Median for all States is 11.7 cents.
The top 10 States using Coal in average charge — 8.4 cents less than the bottom 10 States.
The top 10 States using Coal in average charge — 1.8 cents less than the Median for all States. They charge only 85% of the Median price. They charge — 15% less than the Median price. As reviewed Coal appears to be the cheapest fuel in terms of the ultimate rates passed on to customers.
Based on the above metrics, Hydro is the second cheapest. Wind is the third cheapest, very close behind Hydro.
All three (Coal, Hydro, Wind) come in as cheaper than the Median.
Next, you have Nuclear and Natural gas.
And, as the most expensive by a significant margin is Solar. This does not mean Solar is not an efficient fuel for electricity generation. It is. Instead, it means that the utilities that do use more Solar than other utilities typically charge higher rates or prices to customers.
The above rankings are pretty consistent with the earlier methodology, Scatter Plots & LOESS.
Studying the relationship using linear regression
This method is interesting because it allows to differentiate between the effect of fuel vs. a State on electricity price. A State with inefficient utility operations will charge higher electricity rates than another State with more efficient utility operations. And, as we will soon see the State effect is very strong.
Below, I disclose the summary statistical description of 5 different regressions.
From left to right, I first show the best model from a fit standpoint. As inputs, it includes a combination of both fuels and States.
The second model uses fuels only. This model represents a fuel-pure play to understand the direct impact that each fuel has on the price of electricity excluding the State-effect.
The third and fourth models use only a few States to fit the same State electricity price data. These models uncover the very strong State-effect on electricity prices charged to customers.
Notice that the relevant models include petroleum as a fuel. As mentioned earlier, petroleum is really an outlier. It is mainly used in Hawaii, where it generates about 65% of its electricity. Hawaii has by far the highest electricity price in the US. Both the petroleum and Hawaii regression coefficients are frankly not much more than noise associated with an outlier. However, regression models including these outliers fit the overall data on all the other States electricity prices much better than otherwise.
So, what do the first two models including fuels tell you?
a) They again confirm that Coal is the cheapest fuel with regression coefficients associated with a reduction in electricity prices ranging from — 5.2 to — 8.2 cents per kWh relative to the Intercept or Constant in the models;
b) Wind and Hydro, depending on the models also come in with a reduction in electricity prices relative to the Intercept in the models;
c) Solar, again comes in last as staggeringly more expensive than the other fuels in terms of the resulting prices or rates charged to customers. California is the national leader in Solar. So, when we use the best model that differentiates the impact of fuel vs. State on electricity prices, the model assigns a + 10.5 cents for California’s relative inefficiencies and + 11.3 cents for the Solar fuel itself. When we use the model just using fuels, it assigns + 24.8 cents to Solar alone. These electricity prices mark ups are gigantic when you figure that the US median price of electricity is 11.7 cents per kWh.
d) Natural gas and Nuclear, within those models come in as neutral. This means that their use as a fuel would not cause the price of electricity to deviate much from the Intercepts in the models that come in at 13.3 to 14.4 cents per kWh.
Again, the tiering of the fuels impact on rates and prices is pretty consistent with the other two methods we used (Scatter plots & LOESS, and top 10 vs. bottom 10 comparisons).
What do the two models just using States tell you?
They indicate that the State-effect on electricity prices is enormous. Just using a model with 4 States (California, Connecticut, New Hampshire, Hawaii) can explain more in the variance within the States electricity prices data then knowing the States’ respective fuel mix. You can see that by the models’ adjusted R Squares.
The model with only fuels has an adjusted R Square of 0.58 and a Standard error of 3.6 cents per kWh.
The model with 4 States has an adjusted R Square of 0.62 and a Standard error of 3.4 cents per kWh.
The model with 6 States has an adjusted R Square of 0.72 and a Standard error of 2.9 cents per kWh.
Conundrum as a conclusion
As reviewed, the estimated cost of different fuels to generate electricity (input) have often little relationship with the price that utilities charge to customers (output).
For instance, Solar is associated with the lowest costs on the generating side, but the highest price to customers on the selling side.
Similarly, Coal is associated with a much higher cost than Solar on the generating side, but a far lower price to customers on the selling side.
The State effect plays a substantial role in the above distortions
between economic input (fuel costs) and output (electricity price to
customers). I have studied the particular case of California to better
understand the State effect within this article: What's up with California electricity rates?
THE END