Relating Product Prices to Long-Run Marginal Cost: Evidence from Solar Photovoltaic Modules

doi:10.1111/1911-3846.12319

Lawrence Berkeley National Laboratory (LBNL)’s annual Tracking the Sun report summarizes installed prices and other trends among grid-connected, distributed solar photovoltaic (PV) systems in the United States. 1 This edition focuses on systems installed through year-end 2018, with preliminary trends for the first half of 2019. As in years past, the primary emphasis is on describing changes in installed prices over time and variation across projects. This year’s report also includes an expanded discussion of other key technology and market trends, along with several other new features, as noted in the text box below. Trends in this report derive frommore » project- level data reported primarily to state agencies and utilities that administer PV incentives, renewable energy credit (REC) registration, or interconnection processes. In total, data were collected and cleaned for 1.6 million individual PV systems, representing 81% of all U.S. distributed PV systems installed through 2018. The analysis of installed prices is based on the subset of roughly 680,000 host-owned systems with available installed price data, of which 127,000 were installed in 2018. A public version of the full dataset is available at trackingthesun.lbl.gov. Numerical results are denoted in direct current (DC) Watts (W) and real 2018 dollars. Non-residential systems are segmented into small vs. large non- residential, based on a cut-off of 100 kW. Distributed PV Project Characteristics. Key technology and market trends based on the full dataset compiled for this report are as follows. • PV systems continue to grow in size, with median sizes in 2018 reaching 6.4 kW for residential systems and 47 kW for non-residential systems. Sizes also vary considerably within each sector, particularly for non-residential systems, for which 20% were larger than 200 kW in 2018. • Module efficiencies continue to grow over time, with a median module efficiency of 18.4% across all systems in the sample in 2018, a full percentage point increase from the prior year. • Module-level power electronics—either microinverters or DC optimizers—have continued to gain share across the sample, representing 85% of residential systems, 65% of small non- residential systems, and 22% of large non-residential systems installed in 2018. • Inverter-loading ratios (ILRs, the ratio of module-to-inverter nameplate ratings) have 1 In the context of this report “distributed PV” includes both residential as well as non-residential rooftop systems and ground-mounted systems smaller than 5 MWAC (or roughly 7 MWDC). An accompanying LBNL report, Utility-Scale Solar, addresses trends in the utility-scale sector, which consists of ground-mounted PV systems larger than 5 MWAC. New Features in This Year’s Tracking the Sun • Expanded Discussion of Project Characteristics. This year’s report includes additional trends related to distributed PV orientation, inverter loading ratios, and solar-plus-storage. • Focus on Host-Owned Systems for Installed Pricing Analysis. In order to simplify the analysis and discussion, the report now excludes third-party owned systems from its analysis of installed pricing trends, though those systems are included when characterizing broader technology and market trends. • Multi-Variate Regression Analysis. The report now includes an econometric model of installed pricing variation across residential systems installed in 2018 (see Appendix C), complementing the descriptive analysis. 2 Tracking the Sun generally grown over time, and are higher for non-residential systems than for residential systems. In 2018, the median ILR was 1.11 for residential systems with string inverters and 1.16 for those microinverters, while large non-residential systems had a median ILR of 1.24. • Roughly half (52%) of all large non-residential systems in the 2018 sample are ground- mounted, while 7% have tracking. In comparison, 17% of small non-residential systems and just 3% of residential systems are ground-mounted, and negligible shares have tracking. • Panel orientation has become more varied over time, with 57% of systems installed in 2018 facing the south, 23% to the west, and most of the remainder to the east. • A small but increasing share of distributed PV projects are paired with battery storage, typically ranging from 1-5% in 2018 across states in our dataset, though much higher penetrations occurred in Hawaii and in a number of individual utility service territories. • Third-party ownership (TPO) has declined in recent years, dropping to 38% of residential, 14% of small non-residential, and 34% of large non-residential systems in the 2018 sample. • Tax-exempt customers—consisting of schools, government, and nonprofit organizations— make up a disproportionately large share (roughly 20%) of all 2018 non-residential systems. Temporal Trends in Median Installed Prices. The analysis of installed pricing trends in this report focuses primarily on host-owned systems. Key trends in median prices, prior to receipt of any incentives, are as follows. • National median installed prices in 2018 were $3.7/W for residential, $3.0/W for small non- residential, and $2.4/W for large non-residential systems. Other cost and pricing benchmarks tend to be lower than these national median values, and instead align better with 20th percentile values (see Text Box 5 in the main body for further discussion of these issues). • Over the last full year of the analysis period, national median prices fell by $0.2/W (5%) for residential, by $0.2/W (7%) for small non-residential, and by $0.1/W (5%) for large non- residential systems. Those $/W declines are in-line with trends over the past five years. • Over the longer-term, since 2000, installed prices have fallen by $0.5/W per year, on average, encompassing a period of particularly rapid declines (2008-2012) when global module prices rapidly fell. In many states, the long-term drop in (pre-incentive) installed prices has been substantially offset by a corresponding drop in rebates or other incentives. • Preliminary and partial data for the first half of 2019 show roughly a $0.1/W drop in median installed prices compared to the first half of 2018, though no observable drop relative to the second half of 2018. Those trends are based on a subset of states, consisting of larger markets, where price declines have recently slowed compared to other states. • Installed price declines reflect both hardware and soft-cost reductions. Since 2014, following the steep drop in global module prices, roughly 64% of the total decline in residential installed prices is associated with a drop in module and inverter price, while the remaining 36% is due to a drop in soft costs and other balance-of-systems (BoS) costs. For non- residential systems, a slightly higher percentage of total installed price declines is attributable to BoS and soft costs. Variation in Installed Prices. This report highlights the widespread variability in pricing across projects and explores some of the drivers for that variability, focusing primarily on systems installed 3 Tracking the Sun in 2018. The exploration of pricing drivers includes both basic descriptive comparisons as well as a more formal econometric analysis. Key findings include the following. • Installed prices in 2018 ranged from $3.1-4.5/W for residential systems (based on the 20th and 80th percentile levels), from $2.4-4.0/W for small non-residential systems, and from $1.8-3.3/W for large non-residential systems. • Installed prices within each customer segment vary substantially depending on system size, with median prices ranging from $3.3-4.3/W for residential, from $2.7-3.4/W for small non- residential, and from $2.0-3.6/W for large non-residential systems, depending on size. • Installed prices also vary widely across states, with state-level median prices ranging from $2.8-4.4/W for residential, $2.5-3.7/W for small non-residential, and $1.7-2.5/W for large non-residential systems. • Across the top-100 residential installers in 2018, median prices for each individual installer generally ranged from $3.0-5.0/W, with most below $4.0/W. • Median prices are notably higher for systems using premium efficiency modules (>20%) and for systems with microinverters or DC optimizers. Comparisons between residential retrofits and new construction, and comparisons based on mounting configuration, are both less revealing, likely due to relatively small underlying sample sizes. • The multi-variate regression analysis, which focuses on host-owned residential systems installed in 2018, shows relatively substantial effects associated with system size (a $0.8/W range between 20th and 80th percentile system sizes) and with other system-level factors, including those related to module efficiency (+$0.2/W for systems with premium efficiency modules), inverter type (+$0.2/W for systems with either microinverter or DC-optimizers), ground-mounting (+$0.3/W), and new construction (-$0.5/W). • In comparison, the regression analysis found relatively small effects for various market- and installer-related drivers—including variables related to market size (a $0.2/W range between the 20th to 80th percentile values for market size), market concentration (a $0.1/W range), household density (a $0.2/W range), average household income (no effect), and installer experience (no effect). • After controlling for various system-, market-, and installer-level variables, the regression analysis still found substantial residual pricing differences across states (a $1.5/W range), indicating that other, unobserved factors significantly impact installed prices at the state- or local-levels.

Title: Relating Product Prices to Long-Run Marginal Cost: Evidence from Solar Photovoltaic Modules

Citation Formats

Robust Standard Errors for Panel Regressions with Cross-Sectional Dependence
journal, September 2007

A wafer-based monocrystalline silicon photovoltaics road map: Utilizing known technology improvement opportunities for further reductions in manufacturing costs
journal, July 2013

Assessing the drivers of regional trends in solar photovoltaic manufacturing
journal, January 2013

Intertemporal Cost Allocation and Investment Decisions
journal, October 2008

The Learning Curve and Competition
journal, April 1981

Accrual Accounting, Informational Sufficiency, and Equity Valuation
journal, October 2011

Dynamics of Rate-of-Return Regulation
journal, May 2012

Levelized Product Cost: Concept and Decision Relevance
journal, July 2015

On the Relationship Between Historic Cost, Forward Looking Cost and Long Run Marginal Cost
journal, January 2011

Managerial Performance Evaluation and Real Options
journal, May 2016

Volatility Forecasting Using Financial Statement Information
journal, September 2015

The electricity market game
journal, April 2003

A Critical Overview of the Use of Full‐Cost Data for Planning and Pricing
journal, December 2002

The dynamics of solar PV costs and prices as a challenge for technology forecasting
journal, October 2013

Cost-based transfer pricing
journal, April 2011

Depreciation Rules and the Relation between Marginal and Historical Cost
journal, June 2009

Modeling the Cost and Minimum Sustainable Price of Crystalline Silicon Photovoltaic Manufacturing in the United States
journal, April 2013

Full-Cost Transfer Pricing and Cost Management
journal, October 2016

A model of competition in the solar panel industry
journal, November 2013

Decentralized capacity management and internal pricing
journal, April 2010

Price Forecasting and Strategic Planning: The Case of Petrochemicals
journal, February 1975

Optimal capital policy, the cost of capital, and myopic decision rules
journal, December 1964

Confidence intervals for dependent data: Equating non-overlap with statistical significance
journal, October 2010

Learning and Forgetting: The Dynamics of Aircraft Production
journal, September 2000

The Learning Curve and Pricing in the Chemical Processing Industries
journal, July 1984

How Q and Cash Flow Affect Investment without Frictions: An Analytic Explanation
journal, March 2011

Consistent Covariance Matrix Estimation with Spatially Dependent Panel Data
journal, November 1998

Drivers of cost reduction in solar photovoltaics
journal, July 2015

Learning by Doing and Dumping in the Semiconductor Industry
journal, April 1991

Activity‐Based Pricing in a Monopoly
journal, April 2003

Capacity Planning and Pricing under Uncertainty
journal, December 2002

Crystalline silicon photovoltaics: a cost analysis framework for determining technology pathways to reach baseload electricity costs
journal, January 2012

Title: Relating Product Prices to Long-Run Marginal Cost: Evidence from Solar Photovoltaic Modules

Citation Formats

Robust Standard Errors for Panel Regressions with Cross-Sectional Dependence journal, September 2007

A wafer-based monocrystalline silicon photovoltaics road map: Utilizing known technology improvement opportunities for further reductions in manufacturing costs journal, July 2013

Assessing the drivers of regional trends in solar photovoltaic manufacturing journal, January 2013

Intertemporal Cost Allocation and Investment Decisions journal, October 2008

The Learning Curve and Competition journal, April 1981

Accrual Accounting, Informational Sufficiency, and Equity Valuation journal, October 2011

Dynamics of Rate-of-Return Regulation journal, May 2012

Levelized Product Cost: Concept and Decision Relevance journal, July 2015

On the Relationship Between Historic Cost, Forward Looking Cost and Long Run Marginal Cost journal, January 2011

Managerial Performance Evaluation and Real Options journal, May 2016

Volatility Forecasting Using Financial Statement Information journal, September 2015

The electricity market game journal, April 2003

A Critical Overview of the Use of Full‐Cost Data for Planning and Pricing journal, December 2002

The dynamics of solar PV costs and prices as a challenge for technology forecasting journal, October 2013

Cost-based transfer pricing journal, April 2011

Depreciation Rules and the Relation between Marginal and Historical Cost journal, June 2009

Modeling the Cost and Minimum Sustainable Price of Crystalline Silicon Photovoltaic Manufacturing in the United States journal, April 2013

Full-Cost Transfer Pricing and Cost Management journal, October 2016

A model of competition in the solar panel industry journal, November 2013

Decentralized capacity management and internal pricing journal, April 2010

Price Forecasting and Strategic Planning: The Case of Petrochemicals journal, February 1975

Optimal capital policy, the cost of capital, and myopic decision rules journal, December 1964

Confidence intervals for dependent data: Equating non-overlap with statistical significance journal, October 2010

Learning and Forgetting: The Dynamics of Aircraft Production journal, September 2000

The Learning Curve and Pricing in the Chemical Processing Industries journal, July 1984

How Q and Cash Flow Affect Investment without Frictions: An Analytic Explanation journal, March 2011

Consistent Covariance Matrix Estimation with Spatially Dependent Panel Data journal, November 1998

Drivers of cost reduction in solar photovoltaics journal, July 2015

Learning by Doing and Dumping in the Semiconductor Industry journal, April 1991

Activity‐Based Pricing in a Monopoly journal, April 2003

Capacity Planning and Pricing under Uncertainty journal, December 2002

Crystalline silicon photovoltaics: a cost analysis framework for determining technology pathways to reach baseload electricity costs journal, January 2012