Deconstruction of a bad study – EV fuel pricing

One of the current challenges in the field of Journalism is that there are currently 6 PR staff for every 1 Journalist, and the PR staff are much higher paid. Which means that a lot of reporting on more technical areas ends up being mostly taking an existing press release, doing an interview with the authors, and running it as a story. Especially for highly technical items (like studies / reports), the Journalists may not have the expertise to really pull them apart with a critical eye to ensure they hold water.

Such was with this Detroit Free Press Article – “Study compares electric vehicle charge costs vs. gas — and results were surprising“:

Last year, Patrick Anderson went electric: He got a Porsche Taycan EV in dark blue.

Anderson, who is CEO of East Lansing-based economic consulting firm Anderson Economic Group, loves the zippy acceleration and “exciting” features the car offers. He also gets satisfaction in knowing that driving an EV benefits the environment, he said. 

But Anderson’s joy comes with a dark side.

“They are a wonderful driving experience. But at the same time, they’re an enormous burden in time and in energy in finding chargers and getting them charged,” Anderson said. “And you’re not really saving much in terms of charging costs … you may be paying more.”

Costs to drive an EV compared with a gasoline car are detailed in a report Anderson Economic released Thursday called “Comparison: Real World Cost of Fueling EVs and ICE Vehicles.”

This seemed really odd, because it goes in the face of a lot of existing studies which show how much cheaper EVs are. Enough so that I went into the study to figure out why.

Finding the Critical Assumption

The critical table in the study is here, where it shows EV cars cost more to fuel. Often by a factor of 2 (highligted in green). But the reasons why that math comes out like that is largely because of what’s in red.

The key assumption of the study is that EVs will be charged outside the home for 70% of their electricity. And that electricity will cost 3x the residential rate. They create a single “mostly at home scenario” which still assumes that 40% of electricity comes from outside the home.

This is a pretty key assumption, that has no justification. The DOE typically assumes that more than 80% of charging happens at home. My own personal experience with 14,000 MWh of energy transfered to our 4 year old Chevy Bolt, is that at least 97% of our charging was at home. And all our charging out of home was actually at for pay L2 chargers which were the same energy price at home (largely at work and during a conference to Troy).

If you replicate this with those numbers, you find the EV costs are in line with their ICE cars. Which is still a surprising result. And that’s where it’s worth looking at all the other biases put into the study.

Lot’s of other assumptions and biases

One other thing was strange on the initial assumption. EVs are divided into Mid Cost (Bolt, Model 3), and Luxury (Porche / Model S), however only the Luxury case showed a 60% at home charge case. That seems like a really big missing case. Why?

If it had been included, it would have broken the streak of every EV scenario coming out more expensive than every ICE scenario. Which seems to be the main goal of the study. This scenario is conspicuous in it’s absence.

Let’s look at other assumptions made and what bias they bring to the table:

  • 12,000 miles per year. The national average is closer to 14,000, and in the state of Michigan, it’s closer to 15,000. As you increase the miles per year EVs cost less. 12,000 is a justifiable number, but it definitely biases towards ICE.
  • Per year state fee on EVs. These range from $0 – $200 depending on the state. $200 was chosen as the number, because that’s the Michigan number. Even though mileage wasn’t chosen for Michigan. Again biases towards ICE.
  • There is a whole set of “dead head” calculations which are used saying you have to drive around looking for chargers, and the number of times you have to go out to do this per month with an EV. Based on the listed fuel economy the number of trips is quite high. It suggests an average of 87 miles per Fast Charge added to a Bolt / Model 3 (which have 240 mile ranges) or 117 miles per Fast Charge for something like a Model S with 250 – 300 mile range. Which seems, odd. Also, even though all other numbers are averaged out to yearly numbers, for some reason the 60% at home scenario actually assumes charging only 100 miles per fast charge session.

Cost of Chargers

There is one good insight in that most EV chargers want to charge at home, and that does mean buying a Level 2 charger and having it installed. They estimate that cost at $1600 with professional installation. They don’t include the 30% tax credit for doing so (which comes and goes, but again that’s an ICE bias). $1600 is probably a little on the high side, I think it cost us more like $1200 all in. They also assume the lifespan of the charger is 5 years, and amortize it over that. That seems extremely low, and again is an ICE bias.

The really bonkers assumption they make though is that you have to buy a Level 1 charger. Those come as standard equipment with every EV. There is really no justification other than an ICE bias to put that in.

Cost of Fueling?

The study authors justify many of these items, as well as not including the costs for an ICE vehicle during this same 5 year window (scheduled maintenance, oil changes, breaks, transmission fluid, etc). They claim annual registration fees are fueling, as well as permanent home infrastructure like a charger.

And this is where it is really clear the point of the report was to generate a very specific result. If you squint hard, pile on some biases, and assume everyone is refueling EVs like ICE cars, you can create a scenario where one part of owning an EV is more expensive than an ICE car.

But, it will still be cheaper to own. Even in the flawed study. And once you start correcting for lots of the flaws, you end up with something that looks a lot like other analyses.

A Real Study would have been…

A real study would have been to note there are some extra costs for EVs outside of just home electricity, like home charging infrastructure, commercial fast charging costs, and state policy creating EV fees, and look at cross over points. Under what circumstances do these extra fees hurt EV adoption. What policies should be in place to less them.

For instance, the EV annual fee to replace missing gas tax only makes sense if you believe that every EV just causes the same damage to roads, with no other benefits. When every EV replacing an ICE car on the roads reduces fine particulate pollution (from both exhaust and break dust) which directly decreases health care costs for the state. There is a big case to be made that these fees should be eliminated on those grounds (or prevented from being passed in states where they don’t yet exist).

What policies ensure homes are EV ready. What policies are needed to handle the economic problems of fast charging (which mostly is about how commercial power tariffs work). These are all good questions to ask. None of them were.

And this is why when you see a study or report come out from somewhere, don’t assume it’s correct out of the gate. Always bring a critical eye, especially to reports that claim to bring results that are very different from the existing literature.

In order to play with the data more, I built a google sheet that plugs in the existing study, and adds scenarios with higher rates of at home charging, and eliminates the per year EV fee. You can make a copy and adjust some of the inputs to see how it changes things. Though remember that even this is still an incomplete picture, as it doesn’t include the maintenance costs.