Last week I discussed Ivan Penn’s New York Times article on electricity pricing. This week Jim Bushnell contributed to the discussion by observing the myriad factors that have made California’s electricity prices high, and that if you remove California the price increases in other restructured states don’t look that different from vertically-integrated states. Penn’s article draws on a recent working paper from Alexander MacKay and Ignacia Mercadal, Deregulation, Market Power, and Prices: Evidence from the Electricity Sector.
The MacKay and Mercadal paper addresses a question of ongoing interest in electricity economics and regulatory policy: to what extent has regulatory restructuring changed price outcomes, and to what extent is market power a factor in those prices? Regulatory restructuring that occurred in several states in the 1990s was a heterogeneous mix of institutional changes, but was in all cases motivated by nuclear construction cost overruns and other cost increases that stirred public and policymaker interest in rivalrous competition as a disciplinary force to counteract some of the political economy pitfalls of utility regulation. My go-to resource to read on this history is the excellent Richard Hirsh, Power Loss.
MacKay and Mercadal assembled a detailed (annual, 1994-2016) dataset at the utility service territory level that includes both utility own generation and utility procurement contracts; the inclusion of procurement contracts in the analysis is an original contribution and should be a much larger focus of the analysis. They use a difference-in-differences matching estimation to benchmark changes in prices, quantities, and costs for utilities in restructured states relative to vertically-integrated states, and find higher markups over costs in restructured states. They then attempt to infer that those higher markups are due to supplier market power (in the face of inelastic demand) by constructing a Herfindahl-Hirschman Index (HHI) estimate at the state level. They also present a case study of Illinois, a restructured state, compared to two vertically-integrated neighboring states, Missouri and Iowa, and discuss other factors that could be influencing their markup estimates.
The matching estimation is the strongest portion of the analysis. They apply this method here based on its previous use in Deryugina et al. (AEJ: Applied 2020), which was an examination of municipal adoption of retail customer choice aggregation in Illinois. In that paper the fact pattern and the institutional changes were more straightforward than the broad and heterogeneous sweep of regulatory restructuring, so while the empirical analysis is well done, the inferences to draw here are not as clear because of the additional assumptions the authors had to make in this more complicated setting.
This analysis is ambitious because the subject matter is particularly complex. Over the course of a decade in the 1990s and early 2000s, 17 states and the District of Columbia (which the authors apparently exclude from their analysis, according to Table A1) implemented regulatory restructuring that could have included a subset or all of the following institutional changes:
reduced entry barriers to participation in an organized wholesale power market,
required membership in a regional independent system operator and transfer of transmission operation decisions to that ISO,
mandatory utility divestiture (unbundling) of most or all of their generation assets, and
reduced entry barriers for retail energy providers for some or all categories of retail customers.
In all restructured states the incumbent utility retained the regulated distribution functions of delivery (poles, wires, transformers, and substations), retail metering, and the long-term planning associated with regulated delivery. They also continued to own transmission as a regulated asset. Three states, Virginia, Oregon, and Michigan, placed stronger limitations on their restructuring than the other states, with Oregon implementing wholesale market competition for only a few industrial customers but no general wholesale or retail competition, Virginia implementing only wholesale competition, and Michigan implementing retail competition primarily for industrial customers. The restructured states in the Northeast, Mid-Atlantic, and Midwest allowed retail competition with extended phase-in periods of residential rate caps (Pennsylvania's was a decade!) and retained incumbent default service for residential customers. Those default service contracts require the incumbent regulated utility to procure energy, typically taking the form of rolling three-year tranches of firm procurement contracts with independent power producers (IPPs), and the utility passes through those energy costs to their default service customers.
And then there are California and Texas, outliers in the restructured space in opposite directions. After their failed restructuring in 2001 California rescinded retail competition and returned considerable decision-making to within the regulated utility, through their long-term procurement process requirements (after having made long-term contracts illegal 1996-2000). In contrast, Texas is the only US state to have fully quarantined the still-regulated natural monopoly T&D wires utilities, with a competitive wholesale market upstream and myriad robust retail competitors downstream, even for residential customers (Kiesling 2014). Texas was also unique in its "price to beat" formulation of the phase-in of retail competition for residential customers (Kiesling 2009).
This institutional complexity suggests two conclusions. The first is that the categorization of all 17 of these states as "deregulated" is excessively coarse, so coarse that the category lacks good meaning. The authors should discuss this institutional diversity and provide a justification for why lumping all of these restructured states into a single category makes analytical sense. The second is that they should abandon the word "deregulation", which is a meaningless and contested word in this context, and should instead adopt the nomenclature of "restructuring" that academic and policy researchers and policymakers have adopted since the early 2000s. Even if they want to retain the definition that that they have adopted that deregulated = allowing market-based prices (p. 11), some of these states have only adopted wholesale markets, some states have retrenched and limited market-based prices, some states only allow retail markets for some types of customers, and some states retain high entry barriers to rivals in retail markets through incumbent default service. In that sense the definition lacks meaning, and the persistence of regulatory requirements in wholesale and retail markets means that these states are restructured, not deregulated in any meaningful sense.
I have three suggestions for the authors, each of which generally amounts to considering an alternative hypothesis in more depth than they do in the current draft:
focus more on the procurement and renewable PPA contracts and their terms,
think about whether the analysis (the markup and HHI estimates especially) accounts for changes in T&D wires charges, and
take the institutional details into deeper consideration when thinking about the sources of market power.
A large original contribution of this work is its inclusion of data on bilateral long-term contracts, and the authors should take greater advantage of the rich information in them. The paper could do more to dig in to the richness of the bilateral contracts; as currently written it focuses on their origins and the fixed-price nature of their terms. It also does not describe whether the contracts in their dataset are only utility contracts or if they also include retail energy provider contracts; for example, Direct Energy is a retail energy provider serving industrial, commercial, and residential customers, and to serve its customers probably has long-term contracts with IPPs. Does the dataset include those contracts?
In implementing restructuring, utilities generally sold their generation assets to IPPs, sometimes as entire fleets. The authors interpret that transaction as maintaining concentration of ownership, which would introduce high market concentration in restructured states to the extent that utilities (now the buyers in energy contracts) signed multi-year power purchase agreements (PPAs) with the IPP to whom they had just sold their generation fleet. But they don't provide the institutional detail contained in the actual contracts, so it's hard for a reader to assess how prevalent this practice was in the mid-1990s – did most utilities really sell their generation to an IPP and then contract with that same IPP at a fixed rate? Providing some data on the behavior would be informative. And how were the terms negotiated? Utility procurement contract terms would have to be approved by the state PUC, so if the utility and the IPP negotiate a fixed price with a substantial markup, that price had to go through regulatory approval. The paper does not discuss the regulatory decision-making process leading to these contracts. Here the alternative hypothesis is the indifferent buyer/inelastic demand of the utility when they can pass through the energy cost in their incumbent default service, so the analytical emphasis should be on the institutional framework that shapes those incentives to utilities and enables such markups.
Implementing renewable portfolio standards (RPSs) through renewable PPAs is another contractual factor that's relevant here, particularly since their estimates suggest an increase in markups post-2005. Do the contracts in the dataset include renewable PPAs to fulfill state RPS regulatory requirements, and what were their terms? There's a considerable academic literature as well as an ongoing policy discussion of how the early renewable PPAs locked in fixed prices, usually in 20-year contracts, but then post-2010 for wind and post-2015 for solar, production costs fell and renewable energy companies were able to offer more attractive terms compared to the earlier PPAs. Does that factor in to the increase in markups? Here the alternative hypothesis is the RPS regulatory requirements leading to markups.
Another important alternative hypothesis is the effect of transmission and distribution wires charges on the retail price. The paper is largely silent on the role of T&D charges, on how transmission charges are reflected in wholesale prices in restructured states while T&D charges are bundled in vertically-integrated states, and on how T&D charges have changed over their time period. As Ivan Penn noted in his New York Times article last week, utilities in restructured states have been investing more in within-state transmission projects, which increases the wires portion of customer bills. The paper does not discuss this change, nor does it mention the contentious FERC Order 1000 that was intended to increase regional transmission investment but has led to a perverse incentive for utilities to invest more in-state, even when the marginal benefit of regional transmission is higher (say, for interconnecting renewable projects from other states). What if the estimated markups they observe are all due to wires investments and charges? The wires are still part of the regulated footprint, even in restructured states, so if wires charges are the source of higher prices that's not a direct consequence of restructuring.
My third suggestion is to encourage the authors to take more of the institutional details into consideration when evaluating the potential consequences, and causes, of market power arising from the contractual form that restructuring took. The current analysis uses their HHI calculation at the state level as an indicator of concentration and the ability to exercise market power, but this indicator has flaws. Why calculate the HHI at the state level, particularly in states where utilities now had access to a larger multi-state wholesale power market (e.g., PJM, ISO-NE) as well as multiple IPPs? The authors are making assumptions about the market definition and the nature of the contracts that have to be better substantiated in the paper. Their HHI calculation is combining dynamics of being in a larger market for 15% of transactions and being in contracts for 85%, so what's the relevant market/what's the correct denominator? It's not the state.
In general, the paper's argument abstracts too far from institutional details and the regulatory decision-making process to enable the authors to make solid inferences about the sources of market power in procurement contracts. Their combination of market transactions and contract transactions obscures important institutional details, such as the oversight provided in organized wholesale markets by independent market monitors. By abstracting from institutional details, the analysis does not address the effects of ongoing regulation (incumbent default service, RPS, wires charges) and the failure to quarantine the regulated monopoly in all restructured states except for Texas.
Focusing the analysis more explicitly on utility procurement contracts and renewable PPAs and retail energy provider contracts, and their dynamics, would strengthen the paper by focusing on the types of transactions that are more likely to be candidates for the exercise of market power. They could motivate the question, for example, by segmenting the data into market transactions and contract transactions and comparing the markup trends across the two, and then focus on the contracts. Bringing in the considerable insights from the extensive contractual literature in institutional and organizational economics would contribute to the theoretical and empirical foundation of an analysis of these generation contracts. A good place to start is Bajari and Tadelis, who have several papers combining theory and empirical application (RAND 2001, AER 2002, IJIO 2012). I look forward to more work from MacKay and Mercadal as they pursue this project.
Fascinating read. This actually helps me think more clearly on a project I am working on in the car insurance market.
Thanks, Lynne. You are absolutely right. A $6 slice of pizza may be worth $10 to me on any given day. (N.B. I do not pay $X for the flour, $Y for the tomatoes, and $Z for the energy that fires the ovens. I buy a well-defined package of food services that provide 1) nutrition, 2) convenience (time and place of consumption), 3) perhaps status (the new place to hang out), etc. Further, I cannot change the efficiency of nutrition conversion very much. My digestive system is more or less fixed.) With regard to energy, I would suggest that a $0.20 per kilowatt-hour of energy is not merely a imperfect measure of value. No one actually wants a unit of energy. It's value is to power one of a hundred end-uses in a home. Unlike digestion, each end-use could be more efficient in the conversion from commodity to service. The few kWh I use each year to recharge my mobile phone is VASTLY underpriced. (It's about $1 per year and I'd pay a lot more if I had to.) The huge number of kWh I use each year to cool my Houston home are overpriced because the Texas Legislature, PUC of Texas, and market participants have forgotten what matters. They do not want to nurture the institutions necessary to create competitive services that will lower costs. From a data perspective, retail electricity prices are all we have. From an economic theory perspective, I don't think they make any sense. Pizza eaters want pizza. Electricity consumers do not want electricity -- they want hot showers and cold beer. ... I am not suggesting that I know how to do better. I am suggesting we must try. ... I look forward to reading more!