Elsevier

Applied Energy

Volume 283, 1 February 2021, 116266
Applied Energy

The impact of wind, solar, and other factors on the decline in wholesale power prices in the United States

https://doi.org/10.1016/j.apenergy.2020.116266Get rights and content

Highlights

  • Annual average prices declined by $19–64/MWh between 2008 and 2017 in the United States.

  • Estimate counterfactual prices by changing a factor at a time in fundamental model.

  • Dominant driver of decline in average wholesale price is fall in natural gas prices.

  • Total impact of wind and solar on average prices is below $3/MWh.

  • Isolating the impact of individual factors is limited by non-linear interactions.

Abstract

Across multiple organized wholesale power markets in the United States, annual average prices declined by $19–64/MWh between 2008 and 2017 while retirements of thermal power plants accelerated. Several prominent changes over the last decade are often discussed as contributors to this decline in prices. These include growth in wind and solar, a reduction in the price of natural gas, and weakened load growth. Here we construct a fundamental supply curve model for each of seven organized wholesale market regions and use counterfactual simulations to assess the degree to which wind and solar—among other factors—have influenced wholesale electricity prices. We find that growth in wind and solar since 2008 reduced average annual wholesale electricity prices by less than $3/MWh. In contrast the decline in natural gas prices reduced wholesale prices by $7–53/MWh, depending on the region. This suggests that recent thermal-plant retirements in the U.S. are primarily due to low natural gas prices, not growth in wind and solar. Fully isolating the impact of individual factors, however, is limited by non-linear interactions between factors.

Introduction

Across the organized wholesale markets in the United States, average annual wholesale prices at major trading hubs declined by $19–64/MWh between 2008 and 2017. Oft-noted causes include the steep reduction in natural gas prices, the rise of variable renewable energy (VRE, inclusive of wind and solar), and moderating load growth [1]. One consequence of the reduced prices has been growth in thermal-plant retirements [2], [3]. The change in the generating mix in the seven organized wholesale market regions in the U.S. between 2002 and 2016 shows the parallel reduction in coal generation and increase in natural gas and VRE generation, Fig. 1.

Considerable attention has been placed on the impacts of VRE on wholesale power prices. After all, wind and solar power have both grown rapidly, and each has unique characteristics that may have distinctive impacts on wholesale pricing patterns. This literature consistently confirms the so-called ‘merit order’ effect—namely, that the addition of VRE with low marginal costs leads to lower market-clearing prices. And yet, estimates of the absolute magnitude of the effect vary, in part due to various approaches applied to different regions, making comparisons difficult; a summary of the empirical U.S. literature is provided in Table 1, showing a range of historical impacts from $0–12/MWh. Much of the literature has focused on VRE impacts, without consistently considering the wide array of other possible price drivers. The U.S.-focused literature is a subset of the broader literature on the price effect of wind and solar in Europe, with many of the studies summarized by Welisch et al. [4] and Würzburg et al. [5]. Csereklyei et al. [6] find that wind and solar in Australia put downward pressure on prices in the context of overall increasing average prices between 2010 and 2018.

Studies that have considered a wide array of other drivers often use a fundamental model rather than a statistical model. Haratyk [2] develop a simple fundamental model to estimate the impact of changes in natural gas prices, installed wind capacity, demand, and other factors on wholesale prices in the Midwest and Mid-Atlantic regions of the U.S. Kallabis, Pape, and Weber [7] build a simple supply curve model of the price of electricity futures contracts in Germany to determine the drivers of the decline between 2007 and 2014. While the decline is frequently attributed to the increase in renewable generation, they find that emissions prices had a greater impact. Bublitz et al. [8] finds the impact of changes in carbon permit and coal prices were twice the impact of renewable expansion in Germany using both an agent based model and a statistical regression model. Using a fundamental model, Hirth [9], in contrast, finds that growth in renewables was the largest single driver of the wholesale price decline in Germany and Sweden between 2008 and 2015. Hirth explains that differences in conclusions across the European studies stem from differences in the time horizon, geographic coverage, and whether studies focus on futures or spot prices.

We build on the approach described in Kallabis et al. [7] and Hirth [9] to quantify the relative impact of VRE and other factors on annual, market-wide average historical wholesale prices between 2008 and 2017 in the U.S. In contrast to much of the previous U.S. literature, our approach allows for the application of a consistent method across multiple regions, and enables us to compare the influence of wind and solar to many other factors. To do this, we develop a relatively simple, fundamental supply-curve model for all seven centrally organized wholesale markets in the United States: the California Independent System Operator (CAISO), the Electric Reliability Council of Texas (ERCOT), the Southwest Power Pool (SPP), the Midcontinent Independent System Operator (MISO), the PJM Interconnection (PJM), the New York Independent System Operator (NYISO), and the New England Independent System Operator (ISO-NE). Collectively, these seven markets cover more than two thirds of the load in the U.S. Our use of a fundamental model allows explicit representation of non-linear relationships between changes in factors and average prices, a feature missing from much of the previous U.S. literature.

Each of the ISOs runs an energy market and various ancillary service (i.e., reserve) markets, and many also have capacity markets or related resource adequacy obligations. The energy markets clear on at-least an hourly basis, with both a day-ahead financial market and a real-time balancing market to account for changes that occur in near-real-time. We focus exclusively on prices in energy markets at major trading hubs, and primarily on hourly real-time (not day ahead) prices. In order to draw connections between energy prices at major trading hubs and changes to revenues of generators it is important to recognize that the aggregate revenue of a generator participating in a wholesale market depends on a locational marginal price (LMP) at the generator location, which may differ from prices at major trading hubs due to transmission congestion. Total generator revenue also depends on ancillary service prices, capacity prices, and the dispatch of that generator. Moreover, many contracts between generators and loads exist outside of centrally organized wholesale spot markets, in which case the LMPs reflect the grid value of power and establish the opportunity cost of not selling into or buying from the wholesale market but do not necessarily have a direct impact on the contracting parties.

With these supply curve models, we estimate counterfactual prices where one factor is changed at a time. We find that growth in wind and solar since 2008 reduced average annual wholesale electricity prices by less than $3/MWh, a level within the range of the estimates in the literature summarized in Table 1, and on par with several other secondary factors. The primary contributor to the reduction in wholesale prices is the decline in natural gas prices, which, depending on the region, drove prices $7–53/MWh lower over this same period.

Section snippets

Quantifying the relative impact on average wholesale prices

Building on the approach outlined by Kallabis et al. [7] and Hirth [9], we use a fundamental supply curve model, described below, to estimate the change in annual average wholesale prices from changing one factor at a time. In particular, we compare modeled annual average prices when all factors are set to their 2017 levels to counterfactual annual average prices when changing one factor at a time to its 2008 level. For example, we estimate the impact of growth in wind over 2008–2017 on average

Supply-Curve model validation

To validate the supply-curve model, we compare the wholesale prices from the model to actual wholesale power energy prices from major trading hubs in each region. The major trading hubs used for the actual annual average wholesale prices in the real-time market are listed in Table 3. Day-ahead prices from the same hubs are used for years in which the day-ahead market existed. Some of the markets have seen major design changes between 2008 and 2017. In particular, several of these markets did

Discussion

The finding that the reduction in natural gas prices was the primary contributor to the fall in wholesale electricity prices since 2008 is consistent with an emerging literature that has similarly focused on decomposing factors impacting average wholesale prices in the United States, albeit generally focused on a smaller set of possible drivers and a subset of regions. For example, Jenkins [12] estimated the relative impact of different drivers for wholesale price reductions from 2008 through

Conclusions

Wholesale power markets in the United States experienced several major shifts over the past decade, with growth in wind and solar, a steep reduction in the price of natural gas, limited growth in electrical load, and an increase in the retirement of thermal power plants. Building on related work, this article has assessed the degree to which growth in variable renewable energy has influenced wholesale power energy prices in the United States, not in isolation but in comparison to other possible

CRediT authorship contribution statement

Andrew Mills: Methodology, Software, Writing - original draft. Ryan Wiser: Conceptualization, Supervision, Writing - original draft. Dev Millstein: Writing - original draft. Juan Pablo Carvallo: Software, Methodology. Will Gorman: Software, Methodology. Joachim Seel: Resources. Seongeun Jeong: Resources.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by the U.S. Department of Energy (DOE) under Lawrence Berkeley National Laboratory Contract No. DE-AC02-05CH11231. We especially thank our many DOE sponsors for supporting this research. Those sponsors played no role in the design or execution of this research, but did offer useful feedback on earlier drafts of the resulting manuscript. We also appreciate the feedback provided by a number of external advisors to and reviewers of this work.

References (28)

Cited by (25)

View all citing articles on Scopus
View full text