Electricity is an increasingly complex industry in the midst of transition to renewables and decarbonization. Using my 25 years’ experience as an engineer, policy analyst, and academic, I help my consulting clients think through their toughest technical challenges and formulate their best business strategies.
I am pleased to report that my last post (“Renewables: it takes a portfolio”) received the most comments ever! In that post I discussed how to think about constructing a least-cost portfolio of thermal generation, storage, and demand response to complement renewables. In this followup, I would like to respond to attorney Dan Watkiss, who commented that “your economic analysis fails to account for environmental and social externalities — you’re not alone in this very serious flaw in current energy economic analysis.” He’s right. We need to include the environmental and social costs of energy production. What I’d like to do here is to expand my previous energy economic analysis to include environmental externalities. How can we account for the environmental cost of carbon dioxide emissions in designing a least-cost portfolio? I am now expanding “cost” to mean not just capital and operating costs but also the cost to our environment.
There is a simple fix, and it’s not new. It’s the idea of a price on carbon. As many economists have noted (see, for example, the extensive discussion by Professor Robert Stavins of Harvard), the most straightforward approach to pricing carbon is either a carbon tax or a cap-and-trade mechanism. In either case, a price is charged to the carbon emitter for the negative impact on the environment. Many countries now have a price on carbon, and Australia had a price on carbon from 2012 to 2014. In the case of a carbon tax, if we can estimate the marginal environmental cost of carbon dioxide emissions, then we can set a price on carbon emissions equal to this marginal environmental cost. In the short term, this forces fossil generators to pay the costs of polluting the environment. The higher their operating cost, the less competitive they are in the electricity market, the less electricity they’ll sell, and the less profit they will earn. In the long term, generators will be incentivized to invest in carbon-reduction technology, such as carbon capture and sequestration, or exit the industry.
Let’s more carefully compare the incentives provided by a price on carbon emissions versus not pricing emissions. First, on the demand side: without a price on carbon that reflects the underlying cost to the environment, we will tend to consume too many carbon intensive resources. Professors Severin Borenstein of the University of California, Berkeley, and Jim Bushnell of the University of California, Davis, have found that volumetric charges (per kWh prices) to consumers to recover the fixed costs of various policy measures will sometimes result in retail prices that are too high in US states such as California, but which are too low in several other US states. In locations where retail prices are too low to fully reflect the social cost of emissions, consumption can be expected to exceed socially optimal levels: people are using too much electricity compared to other fuels. For example, if retail gas and electric prices do not correctly reflect emissions costs, then shifting from gas to electric heating in a coal-dominated electricity system might result in higher emissions, and higher overall costs considering capital, operating, and emissions costs than if retail prices correctly reflected the social cost of emissions.
On the supply side, the lack of a price on carbon will tend to result in an inefficient mix of generation: too much generation from resources that emit carbon dioxide, and too much capacity of such high emission resources. Again, this means that the overall costs including capital, operating, and emissions will be higher than they should be. For a concrete example of the implications for generation dispatch, see exercise 7.2 in my “Locational Marginal Pricing” course (from slide 74 onward). With a price on carbon, market forces will tend to bring the industry toward the goal of minimizing overall capital, operating, and environmental costs.
However, the political reality is that governments are loathe to impose a price on carbon, particularly where the fossil fuel industry is influential. In Australia, for example, the Federal Liberal government’s net-zero mantra is “technology not taxes.” It is betting that as-yet-unknown advances in technology will get the country to net-zero by 2050, without any need for a price on carbon. And, perhaps even more importantly, without hurting the influential coal industry. Does this “technology not taxes” strategy, repeated by representatives of the Australian government, including Prime Minister Scott Morrison, stand up to scrutiny? Will it get Australia to net-zero without a price on carbon dioxide emissions?
The short answer is no. For one thing, the technologies will not be deployed without governmental incentives to build the technology or disincentives against emissions. It’s common sense—why would anyone invest any money in new technology that comes without any added profit unless they are forced to? (For a smart analysis of all the reasons why “technology without taxes” is unreasonable, see these articles by the Sydney Morning Herald’s Economics Editor Ross Gittins: Praying, Net zero, Masterpiece.)
Are there alternatives to a price on carbon to reduce emissions? Given the political difficulties of pricing carbon dioxide emissions in both Australia and the US, could a subsidy, such as the US Production Tax Credits (PTCs), do the job instead? Although PTCs and other subsidies worldwide have helped to spur technology that has reduced the costs of building renewables, subsidies simply cannot get the same results as a price on carbon. There are at least two reasons. First, electricity prices end up being too low, thus incentivizing greater consumption. This may also result in too little generation capacity overall, leading to supply adequacy problems. Second, the price advantage of the subsidy to renewables does not differentiate between the emissions levels of the other resources, causing the wrong mix of supply-side thermal resources. For example, coal and combined cycle gas generation see the same differential price effect given a subsidy on renewables, even though coal generation emits approximately twice as much carbon dioxide as combined cycle gas generation. This gives a relative advantage to coal compared to gas when looked at from the perspective of total capital, operational, and environmental costs. With the same differential price due to the renewable subsidies, there will be too much coal generation relative to natural gas generation.
Now let’s look at carbon sequestration, one of the technologies that will likely be needed to get us to net zero. The Liberal Australian government speaks as if the technology for carbon capture and sequestration is so cheap, or even free, that no governmental mandate or regulation is required to see it implemented. But, in fact, there are significant capital and operation costs for sequestration. Without a price on carbon, and without a regulatory mandate or subsidies, what would motivate a coal or gas fired power station to invest large sums to capture and sequester its carbon dioxide emissions? It is hard to see why any company would do so. Without a price on carbon, carbon capture and sequestration would require mandates or subsidies, or both, because its capital and operating expenses are very high.
Is there a drawback of such mandates and subsidies? Targeting mandates and subsidies to particular market segments will likely mean that cheaper options to decarbonize are overlooked. That is, when any government picks winners and losers, as it does when instituting mandates or offering subsidies, it likely makes the overall costs higher than they need to be. “Technology not taxes” will not result in minimizing overall capital, operating, and environmental costs.
Although the public discussion in Australia is not clear, subsidies to particular segments of the economy appear to be central to the Australian government’s decarbonization plans (see “Morrison’s Tricky Deal” and “Barnaby’s Billions”). Subsidies to one industry must be paid for somehow. Where does the money come from? Subsidies must be funded out of taxes on other parts of the economy. So, the Australian government’s plan would be more properly described as “technology and taxes to fund subsidies.” So, not only does “technology not taxes” not bring us toward minimizing overall costs, it also actually involves increased taxes.
A significant argument against pricing carbon is its disproportionate impact on low- and middle-income earners. How much more will low- and middle-income earners pay with a price on carbon than without? Will this affect their income negatively? Greenhouse emissions vary significantly by country and by person, and the accounting is complicated by the effect of carbon dioxide compared to other greenhouse gases, but we might estimate an average on the order of about 15 tonnes of carbon dioxide equivalent per person per annum in Australia and the US. The marginal environmental cost of carbon dioxide emissions is contentious, but let’s consider an indicative value of US$50 per tonne. Charging for carbon dioxide emissions at this price would cost each person an average US$750. If this was in the form of a carbon tax, then the money would add to other taxes paid to the government.
How does that stack up compared to taxes and subsidies in typical income tax filings? As an example, the US “Earned Income Tax Credit” provides a per capita subsidy that ranges from around US$1500 to US$6700 per year per taxpayer for low- and moderate-income workers. To compensate low- and moderate-income earners for their roughly US$750 annual payments for carbon, we could increase their Earned Income Tax Credit by that amount. Analogous adjustments could be made in Australia and other countries.
I have discussed taxes, but what about technology? The “technology not taxes” mantra is half right. Technology is a necessary driver of carbon reduction, and we need to be aggressive about developing new low emissions technologies and improving energy efficiency. Moreover, some limited subsidies for early-stage technologies can be a great investment if they catalyze cost reductions for subsequent large-scale deployment. For example, the early effect of PTCs and Investment Tax Credits (ITCs) in the US, and other mechanisms such as feed-in tariffs elsewhere, have helped with research and development of renewables and with scaling up the renewables industry, contributing to the astonishing reductions in fabrication costs. This early-stage investment has helped to bring us to the point where the unsubsidized cost of new renewable electricity is now cheaper than fossil electricity.
So Dan, in summary: we can account for environmental externalities in economic analyses. To do this we need a carbon price. If we can achieve that, then we can align everyone’s incentives toward decarbonization without a priori favoring one technology or another, or one industry over another. We need technology and a price on carbon.
Renewables are often touted as being cheaper than fossil generation. Certainly true when we have wind or sun. But when it is not sunny and not windy, we must, by definition, use a more expensive resource. So how do we make sure that the total cost of producing electricity is the lowest possible, considering both the capital (investment) cost and the operating cost, where the latter might include the “cost” of inconvenience to us as consumers of re-scheduling our consumption.
The biggest challenge, then, is to match supply and demand. We all know that renewable production depends on the ambient conditions, varies over time, and does not closely match typical patterns of consumption. This is famously reflected in the California duck curve. In a previous post, I suggested that residential pre-cooling in regions with significant air-conditioning load could match the daily variability of solar production to electricity consumption.
But the fluctuations are not just daily. Variability also presents problems at timescales from the very short-term (minute by minute) to the very long-term (seasonal and longer). Much of the variability has a random character, because of issues such as the weather’s effect on renewable production and electrical consumption. How, then, should we think about matching renewable supply and electrical demand, bearing in mind that we must balance production and consumption in the electricity system at all times?
A first observation is that a portfolio of diverse renewable resources will have lower variability than a single resource. Its mix of resources can also be chosen to best match consumption on average. But that would still leave a discrepancy between renewable supply and electrical demand.
There are many possible solutions for coping with the discrepancy between renewable supply and electrical demand. Generally, I have advocated for demand-side responses to match demand to supply. And now that the cost of chemical battery storage is getting cheaper, batteries also play an important role in aligning renewable production to consumer demand. Where available, hydroelectric resources are useful. (Bill Gross’s Idealab has developed another promising storage technology that lifts heavy weights to store energy.) And I can also see how thermal resources, used sparingly, would help in matching supply to demand, particularly to cope with solar and wind “droughts,” where there might be little wind and solar for several days.
What is the most cost-effective solution? What’s key, I think, is to understand the underlying cost structure of each proposed solution, at least in broad brush. And when considering cost structure, we need to take into account the distinction between the capital cost per kW of power capacity or per kWh of energy capacity, or both, and the operating cost per kWh.
At one extreme, we might consider a solution that is capital intensive, like batteries and some thermal generation, with a high cost per kW of power capacity or per kWh of energy capacity. Such assets are most economical when used very often. Battery storage used on a daily basis to provide ancillary services has been relatively profitable, meaning that the value it delivers can easily justify the expense of the battery. On the other hand, using a battery or a thermal generator only for a once-a-decade condition, such as the extreme weather in ERCOT in February 2021), is likely to be expensive, because the cost is expended for only one occurrence of benefit per decade. In other words, to be cost effective, we need to stack the benefits of batteries. So, if we can use batteries for multiple applications, utilization is overall high enough to justify the cost.
At the other extreme, we might consider solutions that are operational cost intensive, with relatively lower capital cost. They include peaking generation, consumer backup generators, and various forms of demand response that involve interrupting customer service. These solutions are only viable if used very occasionally and for short periods of time.
Some good news on the consumer backup generation front: the upcoming “vehicle-to-home” (V2H) technology, where an electric vehicle battery is harnessed in a microgrid with rooftop solar. Nissan Leaf already has V2H available and next year the Ford F150 Lightning will be available with V2H. V2H is best suited to only occasional use because, obviously, unlike a dedicated generator, you need your car to drive. However, in a winter storm such as Texas experienced in February 2021, I needed power at my house during the blackout and could not even drive on the snow-covered roads, an ideal application for V2H! Such solutions match the rare, but sometimes severe, occurrences of distribution failures and rolling blackouts, stacking this occasional back-up role on top of the daily benefits of having a car. In a highly renewable world, V2H is also a potential solution for renewable droughts.
In between those two extremes, we can also imagine other solutions. Battery technology is improving, pushing its applicability toward uses with lower utilization. For example, some level of battery storage may be appropriate for daily charge and discharge cycles even though it would be insufficient for an extended blackout or renewable drought. Some demand-side adaptations such as residential pre-cooling and industrial demand-response may also be viable on a regular, daily basis. They will not, however, provide for a multi-day event such as the 2021 storm.
If we design a portfolio of solutions with heterogeneous cost characteristics, some with high capital cost and low operating cost, others with low capital cost and high operating cost, and others in between, then adaptation to renewable fluctuations can be accomplished across multiple timescales. If we know the cost characteristics of each proposed item in the portfolio, a “screening curve” provides a good guide to the lowest-cost portfolio. (Tong Zhang has developed a simplified implementation of a screening curve analysis. This version does not consider storage, but it could be used to evaluate the least cost portfolio of generation resources to complement renewables.)
Yes, the least-cost portfolio will involve batteries, but batteries are not the full solution. And neither are demand-side adaptations. It takes a portfolio of resources to match supply and demand. This has always been the case for electricity, but the addition of renewables to supply requires that we need to change how we design the mix of resources.
As has been exhaustively reported, the severe weather event in Texas and surrounding states in mid-February resulted in blackouts across the Electric Reliability Council of Texas (ERCOT) region, including blackouts for all but 90 minutes of a 59-hour stretch at my own home in Austin. Despite some initial claims motivated by animus against renewables, failures occurred across all generation technologies, as well as in gas and water infrastructure, and we can expect that there will be inquiries into the specific ways in which cold caused the Texas blackout.
Much public outcry has been sparked by the federal 2011 report about a less severe Texas winter storm that year. The report recommended weatherization for gas, water, and electricity infrastructure, but they were not made mandatory. Did we fail to learn from the lessons of 2011?
It’s more complicated than that. Before deciding on a solution, we need to be clear about the problem. We must first investigate the facts of the recent storm. Among the things we don’t yet know about weatherization:
- As far as I am aware, we do not know what fraction of electric generation and gas supply in ERCOT actually implemented the recommendations from the 2011 report.
- Electric generators have a large number of sensors, pipes, and other equipment, and we do not know whether the specific equipment that failed in 2011 was the same equipment that failed this time.
- In the gas-fired generation fleet, failures in both the generators themselves and in the gas infrastructure contributed to the outages. Was it gas or electric that was the more significant problem?
Should we have expected private asset owners to weatherize of their own accord? Each such asset owner, as any business owner, runs their operations on a cost-benefit basis. If weatherization improves their expected profit, they will do so. But when the public good is at stake, we can’t always leave risk management decisions to individual actors. In the context of a large “common mode” cause that affects many assets simultaneously as in mid-February, such private actions may be inadequate to appropriately address the risks from the community’s perspective.
When we face risks due to rare events, like winter storms in Texas, whose probabilities may increase with climate change, we cannot depend upon individual asset owners to make risk-averse decisions for the sake of community health and safety. We will likely need to impose regulations and standards to improve extreme weather resilience.
These improvements are likely to focus on weatherization, and they may also include, for example, smart grid technologies to more equitably “spread the pain” of any future blackouts. There may also be market design changes.
But before “firing” on any such actions, we need to “ready” and “aim” by first understanding: 1) what exactly caused the system failure, and 2) what are the costs to avoid these failures or respond to them more effectively. If we truly want to protect our communities, the right actions require thoughtful fact-finding.
Here are several links to selected panel discussions I have participated in and interviews I have given about the Texas blackout. (I will share a comprehensive list next week):
Institute for Operations Research and the Management Sciences (INFORMS) panel with Professor William Hogan (Harvard), Professor Shmuel Oren (Berkeley), Dr. Richard O’Neill (ARPA-E), and Professor Benjamin Hobbs (Johns Hopkins). Click here to view.
Salem Centre for Policy, UT Austin McCombs School of Business panel with Professor Sheridan Titman (UT McCombs), Ms. Bernadette Johnson (Enverus), Ms. Becky Klein (Klein Energy LLC), Professor David Spence (UT Law and McCombs), and Professor John Butler (UT McCombs). Click here to view.
“The Electrical Power Crash Is Just Like a Stock Market Crash,” by Peter Coy. Bloomberg Businessweek. Click here to read.