Electricity generation is a critically important component of the economy and modern life. However, it also generates substantial negative externalities. In particular, electricity generation in the U.S. contributed 31% of U.S. CO2 emissions in 2019 (Energy Information Agency, 2020). Beyond its climate change impacts, electricity generation also emits substan- tial local pollutants that harm human health, with damages for the U.S. estimated at $57.3 billion in 2017 (Holland et al., 2020).

At the same time that research has highlighted the importance of the damages from emissions, we are in the middle of two major transitions in electricity generation. First, over the past two decades, the cost of generating electricity with natural gas has plummeted. This was caused by the pairing of combined cycle generation technology with hydraulic fracturing, which has resulted in gas plants becoming more efficient at the same time that natural gas fuel prices have fallen. Second, the costs of large-scale solar panels and batteries have both dropped over 80% since 2010 (International Renewable Energy Agency, 2020; Cole and Frazier, 2019; Goldie-Scot, 2019). Given these cost declines and the substantially lower pollution externalities from renewables, a transition to renewable energy appears imminent.

These transitions are occurring in a highly regulated environment. Electric utilities’ investment and pricing decisions have historically been regulated as natural monopolies. The 1990s saw extensive electricity deregulation in the U.S. and Europe. In the U.S., this deregulatory push ended with the California electricity crisis of the early 2000s, leaving a patchwork system where different states have substantially different levels of regulatory control.

In its simplest form, regulation of utilities’ pricing and investment sets electricity prices such that firms earn a “fair” rate-of-return on their capital. The canonical model of rate- of-return regulation, Averch and Johnson (1962) (AJ), shows that utilities will respond to this type of regulation by over-investing in capital, suggesting that regulated utilities may have an incentive to over-invest in capacity in these new technologies. However, AJ cannot directly be applied to electricity generation since many of their fundamental assumptions (e.g. elastic demand) do not fit this sector. Moreover, anecdotal evidence suggests that regulated utilities transition more slowly than others: the restructured states of Texas and California lead the U.S. in renewable energy production by a substantial margin (U.S. Energy Information Agency, 2020), and nuclear power plants in states undergoing restructuring disproportionately closed (Davis and Wolfram, 2012).

In this paper we investigate whether the regulatory structure has impeded energy tran- sitions to new technologies and whether it incentivizes utilities to use the existing capital stock inefficiently. To do this, we will develop and estimate a model of regulation with plant investment and exit over the long-run and electricity generation and import decisions in the short-run. Our model builds on the AJ framework in a way that allows us to model the electricity industry. The ultimate goal of this paper is to evaluate how potential regulatory structures could impact the transition to renewable energy.

In the twenty-first century, the fuel used to generate the majority of electricity in the 1 U.S. has transitioned from coal to natural gas. This has led to a 28% reduction in CO2 emissions between 2005 and 2018 and lower local air pollution (U.S. Energy Information Administration, 2018). Two major technological innovations facilitated the evolution toward natural gas. First, combined cycle natural gas generation technology dramatically increased the efficiency with which generators could turn natural gas into electricity. Second, hydraulic fracturing (fracking) has drastically increased the quantity of economically accessible natural gas reserves in the U.S., thereby lowering natural gas prices.

With the recent declines in solar and battery technology prices, we are on the cusp of an- other energy transition. These transitions are occurring in a partially regulated environment, with investment and pricing requiring state Public Utility Commission (PUC) approval in some states but not in others. Even in restructured markets—which have undergone some deregulation—while investment into and exit out of generation are deregulated, decisions on the transmission and distribution of electricity and final consumer prices often still require PUC approval. This makes understanding the policies that facilitated (or hindered) the transition from coal to natural gas paramount.

This paper seeks to understand how the regulatory structure affects energy transitions. In particular, we ask two key questions. First, how does regulation affect the rate at which electricity generation transitions to new technologies? Second, given the existing generation capacity at any point in time, how does regulation affect the use of different generators? To answer these questions, we develop and estimate a model of utility investment and usage under regulation. Our model has both long-run investment and exit and short-run operations components. In the long-run, the regulated utility decides on the amount of coal generation to retire and of natural gas generation to adopt. Within each year, however, the utility makes generation decisions given its investment decisions and exogenous fuel prices. We seek to understand how different regulatory structures would affect energy transitions—both from coal to natural gas generation and, looking forward, to renewable electricity generation.

To understand the current regulatory structure, we adapt the canonical model of regula- tion (Averch and Johnson, 1962, AJ). In the AJ model, the regulated utility is constrained to a maximum return on capital but can price as a monopolist given this constraint. This model is limited in the context of electricity, because electricity demand is highly inelastic implying that the AJ prices are infinite under many functional forms. Thus, our goal is to develop and estimate an updated model of electricity regulation that incorporates key specifics of this sector and that generates predictions that are in line with the observed data. To do this, we start with a model where, each period, regulators deem a set of capital owned by the utility to be effective, representing the “used and useful” standards that state PUCs typically employ. The regulator then reimburses the utility for marginal costs and allows it to earn a rate-of-return on its effective capital. This rate-of-return decreases in consumer electricity rates in order to incentivize the utility to lower costs. The utility chooses its generation and investment decisions to maximize profits in this regulatory environment.

Our model of regulation innovates relative to AJ in three dimensions. First, unlike AJ, the rate-of-return that the utility can earn is declining in the final consumer price of electric- ity. This both recognizes political economy constraints (Wilson, 2019; Hausman, 2019) and prevents the utility from charging arbitrarily high prices, which it could feasibly do under 2 AJ’s rate-of-return formulation by investing in an immense capital stock. Second, in order to model the used and useful standard, we assume that if the utility does not operate the plant sufficiently, it will not receive the rate-of-return on the entire plant capital. This allows for regulation to potentially distort short-run operation decisions in addition to investment decisions, in particular leading to the overuse of old, inefficient generators (Daniel et al., 2020). Third, we adapt AJ to a dynamic setting in order to understand the role of regulation in the adoption of new technologies and in the retirement of existing capital.