Skip to content

Mental Model: Feedback Loops & a Cruel Energy Irony

An important mental model for investors to use is the power of feedback loops. These exist in nature everywhere, as well as some of the best and least-loved sectors in the investment universe. Many internet companies are the beneficiaries of positive feedback loops which help create the “network effect.” The more people that use a particular search engine or social network, the more honed the search results become, or the more useful the network becomes to the user. That’s one of the reasons why search networks are a “winner take all,” type of business model. These feedback loops will continue to reinforce themselves until regulators or anti-trust officials take action.

Feedback loops are also why the companies with the best unit economics in a particular industry often continue to be the market leaders. The more successful these companies are with their customers, the more they can invest in product and service, the more attractive their products & services become, and gaining new customers continues to be easier and easier. You can invert this feedback loop to the weakened competitors that compete against firms experiencing virtuous feedback loops. Physical retail stores vs. Amazon, for example. The more traffic declines, the more they need to either raise price or cut spending. The more they raise price or cut spending, the worse the physical experience becomes. Customer services get outsourced to India, and the overall experience becomes miserable for consumers. Lower profits translate to worse customer experiences or worse products, which continue to reinforce the market share declines.

We find feedback loops a particularly important mental model to use with consumer goods, where economies of scale are particularly important. A company like Telecom Italia, which is generating substantially higher returns on invested capital and profit margins vs. its peers is the only telecom in the country that can afford to invest in fiber. As a result, it will take market share away from other fixed line peers, which will increase its profit margins and ROIC, and allow it to further deploy investments in the country, likely allowing it to have a substantial lead on 5G when its time comes.

But there are far more interesting feedback loops in energy markets which are driving behavior in two very separate directions as a result of the market structure in each segment. One reinforces the use of renewables while the other will dissuade the conversion to renewable energy sources.

Cruel Irony

There’s a cruel irony about the structure of the two major energy markets in the world that will push incentives towards converting the grid to renewables, yet at the same time make electric vehicles less and less likely to become mainstream. We say it’s a cruel irony, because the dream for so many people on the planet would be sustainable energy generation which would in turn fuel electric vehicles. Unfortunately these two realities will be impossible to achieve under the current regulatory structure of the electricity and petrochemical markets.

Because we have a regulated electricity market, one in which the regulators guarantee rates of return for approved capacity additions, infrastructure costs and maintenance expenses are directly passed along to consumers. As more people reach the conclusion that they could actually save money by converting to solar, they push more and more of these infrastructure charges onto a shrinking base of consumers. This in turn makes the regulated electricity increase rates, which further makes the switch to solar compelling for more and more households. As more houses leave the network, electricity rates continue to increase and the incentive to switch over to solar becomes more and more attractive. Unit-level economics of building electricity capacity has always been a small fraction of the actual bills consumers pay to utility companies. Data from the EIA suggests that while companies can often increase electricity capacity at less than $0.01 per kWh, the actual realized priced to consumers in the country has only declined to $0.10 per kWh. The more people that leave the grid, the more these infrastructure maintenance costs gets passed along to its customers. This would be a virtuous feedback loop for renewables and a negative feedback loop for fossil fuels like coal and natural gas.

Exhibit 1: Cost per kWh of various sources of Energy

Key Electricity Costs per kWh
Source: EIA

This will also exacerbate the grid’s balance between renewables during sunny and windy periods as well as having the backup fossil fuel infrastructure to kick on when conditions warrant additional capacity. This standby capacity will be used less and less, yet will still be structured for peak demand, meaning we’ll have significant over-capacity and under-utilization in the electricity markets. This will be passed along to consumers, and will also make night-time electricity more expensive, which is crucial given this is the major period anticipated for the charging of electric vehicles. Fossil fuels will continue to be the major source of night-time electricity generation, which means that electric vehicles will continue to have a larger carbon footprint than fuel-efficient gasoline vehicles.

Aside from Nuclear, which will see almost zero capacity increases in the future thanks to the Fukushima disaster, the cheapest and lowest-carbon emitting way to generate night-time electricity is a combined cycle natural gas plant. This will generate 1.22 pounds of CO2 per kWh which means that after the average transmission loss of 7%, a 301-mile range for the Tesla 85 kWh battery will require energy that generates a carbon footprint of 104 grams / km (the standard unit of measuring carbon output). In contrast, many Toyota and Lexus models are already beating this carbon footprint, and the Prius carbon footprint is already 14.5% lower than Tesla’s Model S. Sure, the Model S has more power than the Prius, but assuming the night-time charging dynamic of electric vehicles doesn’t change, the carbon footprint of fuel efficient gasoline vehicles will always be lower than electric vehicles. In most parts of the country, night-time electricity generation still comes from coal, so the Tesla carbon footprint is even worse than our example. Talk about an inconvenient truth!

Exhibit 2: Carbon Footprints of Electrics vs. Hybrids

Tesla Model S
Toyota Prius
CO2 Output
104 g/km
89 g/km
   Using natural gas-fired electricity
    Using regular gasoline
Range / Capacity
301 Miles / 85kWh


To add insult to injury, the Prius will be sustainably cheaper than any electric vehicle. Because the electricity grid is designed to sustain peak utilization, and is chronically under-utilized as a result of this design, infrastructure costs to consumers are already very high. By switching to rooftop solar, consumers are able to convert a much more expensive power source into a competitive power source by free-riding on other consumers paying for the capacity and maintenance of the grid. Already all over the country, many utilities are asking state regulators to allow them to increase the access fees solar houses pay for backup and night-time access to the grid. Thus, this upwards spiral of infrastructure costs being spread over a thinning base may slow a bit as regulators start forcing grid access fees higher for solar houses. This would lower the incentives for houses to switch over to solar, yet it will only slow and not stop the virtuous feedback loop that exists to support the conversion to renewables.

Oil: $50 the new $90

Gasoline and oil markets have the opposite feedback loop. The more people that convert to electric vehicles and more fuel efficient vehicles, the more demand drops. Given gasoline is unregulated and no longer controlled by OPEC, this means that the price of oil will continue to be suppressed by this lackluster demand environment. Even if electric vehicles never become mainstream, CAFE requirements  demand improvements to fuel economy that will translate into a -1.3% contraction in gasoline demand over the next 10 years. That’s actually with a growing fleet of vehicles driving the same amount of miles on average. China’s fuel economy targets are even more aggressive than the US CAFE requirements, transforming the two largest oil consumers in the world. While Chinese oil consumption will still grow, as its fleet of automobiles will continue growing (even as we believe the auto sales market is poised for a fairly pronounced contraction), American consumption of oil products will decline steadily over the next decade.

Exhibit 3: CAFE Requirements & Resulting Gasoline Demand Changes

Sources: EPA, BTS, GreenWood Estimates using 16 million SAAR for auto sales.

This means that oil prices are unlikely to remain above $50 for protected periods, though spikes to $100 are probably more likely now that OPEC has very little excess capacity in the event of geopolitical turmoil. This represents a major problem for electric vehicles, as the cost parity under the most bullish assumptions require gas to be around $4 a gallon (oil over $130 a barrel).

Models that investment banks have used to justify their bullishness on electric vehicles has always used gas prices around $4 a gallon in order to justify their models showing wide-scale electric adoption by 2020. From the ones we’ve seen, they also assume very little improvement in the fuel economy of internal combustion engines, even though CAFE standards mandate improvements. Plugging in CAFE standards into UBS’s model showing electric powertrain parity, we can see that battery pack costs will need to decline to less than $60 / kWh in order to give internal combustion engines a run for their money. Even Elon Musk thinks he can’t get to lower than $125 / kWh after the Gigafactory is humming along at 100% utilization. If we add powertrain improvements mandated by CAFE just through 2020, the cost parity reduces to $55 per kWh using $2.00 gas and $0.10 per kWh electricity prices (assuming 10,000 miles driven a year).

Exhibit 4: Cost Parity of Batteries in kWh

 Battery Cost Parity

Source: UBS’s Model, GreenWood’s Drivers

The unregulated feedback loop in oil markets mean that if Musk is able to lower battery costs even lower than his goal of cutting them in half to $125 per kWh, the contraction in gasoline demand will mean that even $2.00 is perhaps an unsafe floor for gasoline prices. Thus, the more successful automakers are at improving the fuel economy of their vehicles, the less likely consumers will want electric vehicles. Volkswagen, maker of electric vehicles from both Audi and Porsche, said even at the high-end of the market, Porsche buyers are unwilling to pay a higher price for an electric powertrain option. They’ve claimed that there is zero pricing power for specific powertrain options.

Let’s avoid any confusion, we are big fans of Elon Musk and Tesla, and believe the company is going to destroy the premium segment of the auto industry, and take at least 20% of the demand away from German companies, but we don’t believe the company will be successful making mass-market electric vehicles. Cost parity for electric vehicles requires unrealistic assumptions and is battling a very powerful negative feedback loop that the oil market finds itself in.

The only thing that will hasten the adoption of electric vehicles would be a very conservative government rolling back CAFE requirements and mandated fuel economy improvements, which would strengthen the oil and gasoline demand environment and push prices higher. We wouldn’t rule it out, but isn’t it odd that environmentally-focused voters would best serve their interests by eliminating CAFE requirements? We find it very ironic.

The juxtaposition of a regulated and unregulated feedback loop presents great news for the environment when it comes to the electricity grid, and very bad news for the utility companies who will eventually run out of customers to pass along rate increases to. Yet at the same time, a related but opposite feedback loop will prevent electric vehicles from every gaining scale of any significance.

In our next blog post on mental models and microeconomics, we’ll discuss why we think Tesla will be very successful in its future endeavors despite this negative feedback loop for electric vehicles. No other automotive company has one powerful weapon Elon Musk has, which also happens to represent another important mental model for investors to use. Accordingly, we believe investments in electric vehicles by all other OEMs are doomed for permanent negative returns on invested capital. The Nissan Leaf has been an unmitigated disaster from an ROIC perspective, despite being one of the top-selling electric vehicles in the world.

Feedback loops are very powerful and will continue helping winners and hurting losers until governments and regulators intervene and change policy. Until they do, there will be a very peculiar and ironic dynamic influencing the global conversion to renewable and sustainable sources of energy.

Get Access to Our Research

Would you like to learn how you can access our research on the names mentioned in this article as well as others?


This article has been distributed for informational purposes only. Neither the information nor any opinions expressed constitute a recommendation to buy or sell the securities or assets mentioned, or to invest in any investment product or strategy related to such securities or assets. It is not intended to provide personal investment advice, and it does not take into account the specific investment objectives, financial situation or particular needs of any person or entity that may receive this article. Persons reading this article should seek professional financial advice regarding the appropriateness of investing in any securities or assets discussed in this article. The author’s opinions are subject to change without notice. Forecasts, estimates, and certain information contained herein are based upon proprietary research, and the information used in such process was obtained from publicly available sources. Information contained herein has been obtained from sources believed to be reliable, but such reliability is not guaranteed. Investment accounts managed by GreenWood Investors LLC and its affiliates may have a position in the securities or assets discussed in this article. GreenWood Investors LLC may re-evaluate its holdings in such positions and sell or cover certain positions without notice. No part of this article may be reproduced in any form, or referred to in any other publication, without express written permission of GreenWood Investors LLC.

Past performance is no guarantee of future results.


This Post Has 8 Comments

  1. Very interesting analysis. Thanks for sharing!
    Re. the night-time charging element of the EV/GRID feedback loop, wondering if you have some thoughts on the following variables? and if you think they might reverse the loop?
    1. Energy Storage (for solar systems)
    2. Peak Wind (might be able to absorb some of the night time charging)
    3. Charge at Work (seems like something happening in cities like London)

  2. Crossfire – great point on the peak wind being at night, particularly in the midwest. Only problem being wind’s breakeven still not actually cheap, but with sunk costs into the infrastructure, and the variable costs being zero, in unregulated markets like ERCOT, wind energy often sells for below breakeven costs. If it’s a regulated utility, like Buffett’s MidAmerican / Berkshire Energy, the generation coming from said windmills won’t actually be $0.10 kWh – that being said, it’ll lower the carbon footprint for electric vehicles.

    On energy storage, we’ve obviously been closely following the engineering efforts for cheap capturing of energy, yet Moore’s law hasn’t seemed able to be applied to battery technology, as costs aren’t coming down rapidly. So our view is that wide-scale battery storage efforts would simply add to the utility bill. Or conversely, adding a battery pack in the home also raises the breakeven point at which households are incentivized to switch over. Both of those things make the variable cost of electricity more expensive, and also marginally raise the breakeven cost of switching over to an electric vehicle.

    Your charging at work is a very valid point. In fact, I think the number one reason why people choose electric (besides the fact that after the government subsidy, often times the electric equivalent BMW is actually cheaper than the regular model) is the preferred parking spots they received on electric. Right now when it’s a novelty, these premium spaces are often empty. Try going to the main parking garage in Vail on a Saturday – it’s all full except for the electric vehicle spots. This privilege will dwindle as more and more people adopt electric vehicles. But to your point, if charged using solar, it lowers the carbon footprint of the electric vehicle. At some point in the next 100 years, our grid will be mostly renewable, meaning the lowest carbon footprint alternative will be electric vehicles. That’s not likely happening before 2030 under Obama’s rosiest predictions. Until the government starts regulating carbon like cigarettes, the cost efficiency of a vehicle will still make gasoline cars more compelling for a long while. That’s why the point we made about feedback loops – they will continue until the government or antitrust intervenes.

  3. This is a terrific thought provoking article – so congrats.

    I do wonder about the prospect of technological change in batteries splitting this feedback loop apart also – it might be a decade or more away – but it feels like there is so much research now being done into battery technologies that we might see some step changes in either the storage capacity per unit (which might make it more feasible to have battery costs split among a larger group than a single household unit) or significant cost reductions per unit. If that happens I think the coal market dies a quick death (at least seaborne)

    Your point re Moore’s Law is described in para below from

    “BATTERIES are notoriously hard to improve. Nowhere was this more apparent than at the opening last week, on July 29th, of Tesla’s Gigafactory, a massive battery plant in Nevada. According to its boss Elon Musk, Tesla built the factory because wringing more efficiency out of batteries is far more difficult than optimising the process by which they are made

    1. Matt – Thanks for the comment. There very well may be significant breakthroughs in battery technology, but there will also be breakthroughs in internal combustion engine technology. The fact that the average electric car battery costs well over $15k today and a cheap internal combustion engine costs <$4k means batteries have a long way to go before they become very competitive. I think a lot of auto execs (Ghosn, Kruger, Barra) believe that the market actually wants electric vehicles. We believe this is only the case for a very small group of consumers, and the Tesla success is being driven off the 99% customer satisfaction, which we'll cover in the next blog post next week. Stay tuned and thanks for the commentary!

  4. Good stuff!
    However, I believe that global oil demand is under-estimated (has been for quite a while by most institutional observers) and accelerating this year and beyond (barring a global crisis). So I believe that some of the examples chosen (oil demand and price projections for oil) are at best US centric and not necessarily true – I bet you that oil prices will exceed 65$/barrel by year end and that shortages may potentially be a reality by 2018. Right, OPEC cuts help for now, but with increasing demand… Also, SUV(s)/Truck demand is exploding (I agree hybrid tech is coming there but still…).

    1. Thomas, thanks for the feedback. If you are right, however, battery technology and electric cars charged on a grid with renewables connected (We’re a long way away from there for most of the world, we acknowledge), where incremental or marginal costs of power production is zero, becomes far more competitive. The cruelest irony of them all is that the better the oil patch does, the quicker the patch will meet its demise. We’re going to put out a piece on battery technology soon and unfortunately for oil-drillers, cost-parity is probably less than a decade away at $60-70 oil. That’s actually assuming power at $0.10-0.15 per kWh. In markets like Germany with high renewables penetration, wholesale power prices (not retail) are plummeting, making the transition to renewables for the entire grid even more difficult. So many interesting feedback loops, but if you take a step back from it and look at it at 10k feet – the capacity constraints of the oil market on the global energy market are no longer the only price-setting game in town.

  5. Very detailed analysis. No surprise there.

    While reading, it dawned on me that some of the arguments may well rest on certain assumptions, whose outcomes are rather impossible to predict. To wit:
    (1) battle between distributed and rooftop solar
    I may be wrong, but the negative feedback loop incentivizing/accelerating the shift to solar you suggest assumes the distributed renewables format won’t win out in the end? If it does, you still need base load production via hydrocarbons, unfortunately (or fortunately depending on whose perspective)

    (2) hard to predict the price/value of hydrocarbons. A perusal of its many uses from pharmaceuticals to fertilizers (the soil isn’t getting any less over-worked by the time our great grand kids come around) would lend credence to the impossibility of this forecasting task

    Would love to hear your thoughts on these.

    Thanks for the wonderful work as always.

  6. Interesting post thanks, quick q – using that EIA link you provided I’m getting a CO2 lbs / kWh of 0.99 for 2015 nat gas production in US, which obviously changes the calculus significantly. How’d you get 1.22 number?

    On a more general level, agree with several points but think this is too simplified framework to think about EV and solar generation adoption (granted, dubious that complexity offers better clarity). A few of my thoughts as I read through the article:
    – I’ve typically only seen build costs for different types of generation expressed in terms of cost per power generating ability, not cost per energy generated / consumed, How is that defined? Total lifetime cost of plant divided by utilization x est. useful life? Feel like the comparison to the historical realized price/kWh isn’t apples to apples here.
    – In terms of solar distributed generation being cheaper: haven’t we as a society demanded that generation infrastructure be designed for peak generation? Theoretically shouldn’t solar PV at your home be genuinely cheaper vs the grid only if 1. you are free riding off whatever infrastructure is needed to support peak generation (not a long term market clearing outcome) or 2. your energy consumption habits change or 3. a better way to address peak generation develops? (This all ignores externalities of dirty generation which would change this if properly accounted for)
    – On oil demand reduction as an offset on EV growth – while it’s true the majority of transportation energy consumed is petroleum, the % of petroleum consumed by transportation is much lower %. Don’t really have any knowledge of macro dynamics in other end markets but point is not necessarily a coupled relationship

Leave a Reply

Your email address will not be published. Required fields are marked *

Back To Top