Skip to content

What Drove Solar PV Price Reductions?

A new book explores the precipitous decline.

One of the most remarkable trends in energy economics over the last 50 years is the tremendous reduction in solar photovoltaic (PV) prices. The figure below charts prices on a log scale. In words, it shows that prices in 1970 were about 1,000 times higher than they are currently.

Source: Nemet (2019)

I’ve been on a quest to understand this phenomenon – I’d heard conjectures about scale economies driven by generous subsidies for rooftop installations under the German Energiewende or massive subsidies from the Chinese government, but hadn’t dug into it myself.

It turns out that Greg Nemet, a professor at the University of Wisconsin, recently published a book called, How Solar Energy Became Cheap, which provides a number of insights. And, because there are so many promising energy books to review this year (more on that to come!), I’m going to break my book reviews into pieces and devote this post to Nemet’s book.

Can We Replicate the Solar PV Price Reductions?

Understanding the decline in solar PV prices is a hugely important exercise. If we can replicate the solar experience for energy storage, carbon capture and sequestration, small nuclear reactors or other clean technologies, we’d make a lot of progress addressing climate change. Plus, I’ve had a nagging fear that the current ultra-low PV prices were unsustainable, perhaps reflecting huge subsidies that could disappear with the stroke of a pen.

So, how much of the solar experience actually can be applied to other industries? Was it pure serendipity? For example, did solar PV benefit from advances in the computer industry, which similarly use silicon’s semiconductor properties for microprocessors, that were driven by forces completely outside the energy industry? Or, are there key policies or business practices that can be replicated for other clean energy technologies?

Nemet’s book is based on 70 interviews across 18 countries and also draws on quantitative work that he and others have done for academic publications. It’s a pretty good read, especially given the format – there are footnotes, references, tables and figures, so it’s not exactly a light beach read.

The Case for Replication

At a high level, Nemet comes down on the side of replicability. In fact, the subtitle of his book is, “A Model for Low-Carbon Innovation.” I must admit that I started the book with some skepticism – I was worried that he would be so intent on drawing lessons for other industries that he would ignore the serendipitous explanations. I came away more convinced.

Nemet traces solar PV from the first Bell Lab application in 1954 through to about 2016. He argues that there were essentially four epochs when worldwide PV output was dominated by US, Japanese, then German and finally Chinese production. He devotes a chapter to each country and explores the local demand-pull policies (e.g, Japan’s first-of-a-kind net metering policy in 1998) and technology-push policies (e.g., research and development subsidies, such as the creation of the Solar Energy Research Institute, which later became the National Renewable Energy Lab (NREL)).

Nemet also points out that solar PV benefited from the fact that it was hugely scalable and could be applied across a number of niche markets. The Japan chapter explains how companies like Sharp invested in very small-scale solar cells for calculators and bigger niche applications such as lighting offshore oil platforms.

Solar PV prices really plummeted in the last 15 years, so I found the two chapters on the German and Chinese epochs most interesting. In the chapter on China, Nemet tells the fascinating story of Suntech, the first large-scale Chinese solar module manufacturer. The company was founded by Zhengrong Shi, who spent 15 years as a researcher at an electrical engineering lab at the University of New South Wales in Australia before he returned to China in 2001 with a business plan to build a plant to produce 3 MW of solar modules per year. By 2008, Suntech was producing 1,000 MW per year and had over $1 billion in sales. Suntech went bankrupt in 2013 due in part to its decision to acquire an Italian solar developer later accused of massive fraud, but current powerhouse companies such as Trina Solar, JinkoSolar and JA Solar developed closely on Suntech’s heels.

Shi’s story highlights the importance of academic R&D, as Shi was able to convince German customers that Suntech’s modules were high quality based on his relationship to the university research lab. And, Nemet emphasizes how, “Shi’s experience in the lab [gave] him a broad expertise in PV technologies that allowed him to switch technologies quickly when an opportunity for cost-reductions emerged.”

The story of Suntech and its competitors also emphasizes the doggedness of a number of Chinese entrepreneurs including Shi, who shopped his business plan around China for months before he finally found $5 million in financing. Though Nemet mentions Chinese government policies to invest in renewables to spur demand and says obliquely that, “many of these [solar] firms received substantial funding from the local governments,” I was left with the impression that Chinese solar manufacturing was really launched and developed by scrappy, capitalistic entrepreneurs.

A Series of Incremental Improvements

The scale economies in solar PV are pretty mundane – a bunch of incremental improvements in manufacturing processes, panel efficiency, supply chain optimization, and automation rather than one or two breakthroughs in someone’s R&D lab. For example, Nemet describes how early module manufacturers used second-hand manufacturing equipment repurposed from computer microprocessor plants, but as the industry expanded, suppliers for solar-specific machinery emerged.

He also describes how Shi led the switch from using poly-crystalline silicon to cheaper and more efficient mono-crystalline silicon, which is perhaps as close as we come to an ah-ha insight that drove costs down. It’s splitting hairs a bit, but this last example could be learning-by-doing rather than scale economies as the insight didn’t necessarily require huge volumes of production.

In the end, some of Nemet’s messages are encouraging: the forces of capitalism, nudged by government programs, encouraged entrepreneurs who saw a growing market for solar to invest capital and a fair amount of blood, sweat and tears into fine-tuning manufacturing processes that led to significant cost-reductions. I’m going out on a limb a bit as I’m pretty averse to making forecasts without a lot of evidence to back them up, but I don’t see an obvious reason why the price reductions have been exhausted.

As Nemet points out, though, the decline in solar prices took a long time, so even if other industries can replicate this path, they’ll have to figure out a way to do it much more quickly.

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Wolfram, Catherine. “What Drove Solar PV Price Reductions?”, Energy Institute Blog, UC Berkeley, September 9, 2019,


Catherine Wolfram View All

Catherine Wolfram is Associate Dean for Academic Affairs and the Cora Jane Flood Professor of Business Administration at the Haas School of Business, University of California, Berkeley. ​She is the Program Director of the National Bureau of Economic Research's Environment and Energy Economics Program, Faculty Director of The E2e Project, a research organization focused on energy efficiency and a research affiliate at the Energy Institute at Haas. She is also an affiliated faculty member of in the Agriculture and Resource Economics department and the Energy and Resources Group at Berkeley.

Wolfram has published extensively on the economics of energy markets. Her work has analyzed rural electrification programs in the developing world, energy efficiency programs in the US, the effects of environmental regulation on energy markets and the impact of privatization and restructuring in the US and UK. She is currently implementing several randomized controlled trials to evaluate energy programs in the U.S., Ghana, and Kenya.

She received a PhD in Economics from MIT in 1996 and an AB from Harvard in 1989. Before joining the faculty at UC Berkeley, she was an Assistant Professor of Economics at Harvard.

29 thoughts on “What Drove Solar PV Price Reductions? Leave a comment

  1. Richard McCann regarding your demonstrably false claim that CGNP is relying on a 2010 report. First the CAISO statistics supporting a saturation in California solar installations are current until August, 2019. The references cited are from 2016 and 2019.

    “Turns out wind and solar have a secret friend: Natural gas”, by Chris Mooney, August 11, 2016, The Washington Post, serves as an introduction to:

    “Bridging the gap: Do fast-reacting fossil technologies facilitate renewable energy diffusion?” by Elena Verdolinia, Francesco Vonab, and David Popp, Energy Policy 116 (2018) 242–256,

    • “Richard McCann regarding your demonstrably false claim that CGNP is relying on a 2010 report.”
      Your post refers to a study done in 2010 on the cost of strategies for reducing GHG emissions. That study is obsolete.

    • You’ve made two errors in reading those articles:

      1) This study is about added CAPACITY, not increased ENERGY output. It says little about whether gas generation has increased.

      2) Quote from the article: “Verdolini emphasized this merely describes the past — not necessarily the future. That’s a critical distinction, because the study also notes that if we reach a time when fast-responding energy storage is prevalent — when, say, large-scale grid batteries store solar or wind-generated energy and can discharge it instantaneously when there’s a need — then the reliance on gas may no longer be so prevalent.”

  2. I take exception to solar power apologist Richard McCann’s false statement that zero-emission nuclear power is not dispatchable. Furthermore, baseload nuclear power serves to provide important stability services to the California power grid being subject to the destabilizing effects of large amounts of intermittent solar and intermittent wind. Both German and south Australian power grids have recently been harmed by events connected with a higher penetration of intermittent solar and intermittent wind. Those expensive harms include blackouts and damaged equipment. Grid-intervention events associated with the stops and starts inherent with both solar and wind are increasing for both of these grids.

    CGNP established during the sworn oral cross-examination phase of PG&E witness Frazier-Hampton in CPUC A.16-08-006 that all power systems require baseload power. Specifically, when asked, “Are you aware of any large electric grid, anywhere in the world that operates without a substantial continual supply of electricity from base-load sources?” PG&E witness Frazier-Hampton, who performed their needs analysis, was unable to identify such a grid anywhere. A.16-08-006 Oral Evidentiary Hearing Transcript, April 26, 2017, PG&E, Frazier-Hampton, pp. 946, line 6. Additional details are found in the Appendix, “CGNP’s Summary Responses to other Parties” Dec. 7, 2018.

    Furthermore, the widely-promoted “duck curve” serves to advocate for the inefficient and intermittent dispatch of natural gas fired generation required to compensate for both the intermittencies and poor alignment of the solar power production peak with the late afternoon and early evening load peak. In a previous post, I referenced “Turns out wind and solar have a secret friend: Natural gas,” by Chris Mooney, August 11, 2016, The Washington Post, serves as an introduction to: “Bridging the gap: Do fast-reacting fossil technologies facilitate renewable energy diffusion?” by Elena Verdolinia, Francesco Vonab, and David Popp, Energy Policy 116 (2018) 242–256, The environmental harms identified in those sources remain un-rebutted by McCann

    in this energy policy debate, it is “creative salesmanship” to claim that the increased emissions associated with inefficient and intermittent dispatch of natural gas fired generation yields benefits (except perhaps to natural gas wholesalers.) Furthermore, in its written testimony in CPUC proceeding R.16-02-007, CGNP establishes using CAISO data that “heat rates” for California generators are rising as more and more intermittent solar and intermittent wind are added to the California power grid. For those that are unfamiliar with the term ” heat rate,” it is a measure of efficiency in power production. Rising heat rates mean decreased efficiency.

    • My reply post seems to have disappeared.

      I posted 3 articles that showed the German and Australian power grid problems were unrelated to solar power and that the Tesla batteries has adverted an outage in Australia.

      U.S. nuclear power plants have not load followed on an hourly or even daily basis and are technologically unable to do so. I know of only one plant that cycles on a seasonal basis although there are several with operating costs above springtime power prices. (BTW, if the CAISO has negative hourly prices, why doesn’t Diablo Canyon cycle off in those hours?)

    • This got lost too:
      “Specifically, when asked, “Are you aware of any large electric grid, anywhere in the world that operates without a substantial continual supply of electricity from base-load sources?” PG&E witness Frazier-Hampton, who performed their needs analysis, was unable to identify such a grid anywhere. ”

      That the witness doesn’t know about such a situation does NOT mean that it’s not possible. That’s an incorrect logical leap.

  3. Carl Wurtz raises very important points regarding the non-dispatchability of intermittent solar and intermittent wind. Clearly, the generation resources with the greatest economic value should be dispatchable. While a homeowner may be able to run their dishwasher or clothes washer a few hours later with no economic repercussions. However, a factory, dependent on abundant and reliable power would be forced to shut down during times of no power. Furthermore, the U.S. EPA previously established a “social cost of carbon” that places a significant economic value on the harms of carbon pollution. If electricity markets were rationally-designed the combination of abundance, dispatchability, and zero-emissions would yield the greatest market rewards for nuclear power, trailed by hydropower. As a consequence of 3% fugitive methane emissions during extraction and distribution, dispatchable natural gas and coal would be tied and valued less, and non-dispatchable solar and wind would be valued the least, particularly in California which has too much of both Battery Electric Storage is simply cost-prohibitive.

    • I’ve already addressed the misinformation presented by Dr. Nelson and Carl Wurtz in many previous posts. The first point is that current conventional nuclear plants operating in the U.S. are effectively non-dispatchable. The plants that do have a “variable” output shutdown for an extended period, e.g., Columbia. (And dispatchability is not a highly valuable service based on many modeling exercises. It is certainly much less valuable than either the energy or capacity provided.)

      To say solar hasn’t provided capacity value is to ignore the fundamental reason for the “duck curve”–that solar capacity has met the midday plug load and what remains is evening loads.

      Until small modular reactors (SMR) become commercial, show that they are lower cost and fuel can be disposed of safely, that option is speculative. Nuclear advocates have made many failed promises. We’re done with the “cries of wolf”.

      • If there’s one aspect of renewable energy which can be considered consistent or predictable, it’s the limitless patience of proponents to invent excuses for renewables’ fundamental flaw of intermittency. There are only enough hours in the day to refute them because it’s necessary to refute them. Some well-meaning individuals might take them seriously.

        “To say solar hasn’t provided capacity value is to ignore the fundamental reason for the “duck curve” – that solar capacity has met the midday plug load and what remains is evening loads.”

        Richard McCann, who frequently boasts of his experience in utility regulation, is apparently unaware of the purpose of capacity markets:

        “Capacity represents the need to have adequate generating resources to ensure that the demand for electricity can be met at all times.”

        Let that sink in: “To ensure that the demand for electricity can be met at all times.” Solar and wind energy are unavailable at least two-thirds of every day; availability is unpredictable. Ergo, in terms of ensuring demand for electricity can be met at all times, solar and wind are worthless.

        • As a regulatory witness, I know that the definition of “capacity” complex and context specific. You will find many definitions here for example:

          Notably, “capacity” is defined at “Pmax” which is defined as: “The maximum normal capability of the Generating Unit.”

          Often there are duration limits associated with firm capacity, and discounts for intermittency.

          • Your use of the term “capacity” ignores availability, and is thus entirely outside the context of capacity markets. That there may be discounts for intermittency for wind and solar does not change the fact both are incapable of ensuring demand for electricity can be met at all times. For that purpose, they’re worthless – and any attempt to assign them value constitutes a fraud at the expense of ratepayers.

          • It’s not my term for capacity, it’s the official term used by the CPUC and CAISO. If you were familiar with the calculation of effective load carrying capacity (ELCC) that goes into the NQC, you would know that the definition included availability.

  4. drgenenelson above raises several important caveats to the assumption cheap solar panels translate to lower carbon emissions.

    In a critique of the Energiewende recently, an energy analyst questioned if solar panels were free, whether the expense of land use, of mounting, of cleaning, of backup power, of transmission, of replacement every 25 years, of variability made them worth the drive to the factory to pick them up. A heated debate ensued, one devoid of any conclusive argument it would be.

    That this assumption is even debatable should force us to re-examine the value of dispatchability, of the ability to provide energy on consumers’ terms. When no one can afford to be at the mercy of weather or time of day for electricity, it seems comparing capital or per/MWh prices of dispatchable vs. intermittent energy is pointless.

  5. Dean Wolfram, thanks for the review!

    I’ve got one basic question: Did you actually enjoy reading the book and would you recommend it?

    Sounds like the author makes as good an argument as I’ve heard. Having worked for two solar manufacturing companies, but doing utility scale project development to place he panels rather than help on any aspect of manufacturing, my perspective had been that the industry was built on three pillars: progressive policy making, economies of scale and lots of investment. Policies in Germany, Ontario and California drove solar manufacturing and utility scale development to make and install as many modules as possible. Germans wanted to fight climate change and control a newly meaningful manufacturing sector. California wanted to fight climate change and employ electricians and contractors and encourage its homegrown solar economy. Of course, Chinese entrepreneurs supported by government grants (loans needing no or limited repayment) were able to swamp the Germans and everyone else just about (notable exceptions include First Solar and a few others). The Chinese government wanted to control the industry like they did with wind and employ people.

  6. While this is an interesting post, I believe it ignores classical macroeconomics,. i.e. supply and demand. California has been the state with the greatest amount of commercial photovoltaic (PV) installations of any state. However, there is a saturation in demand as solar now causes significant negative pricing and/or curtailments. Here are some CAISO annual PV commercial installation statistics (megawatts).
    2015 7,861
    2016 10,157
    2017 10,818
    2018 11,960
    2019 (Aug) 12,072

    Last year, California ratepayers paid about a billion dollars to have ratepayers in adjacent states take California’s excess solar power. Apparently, CAISO reset the payment rules to PV owners, so they are curtailing production much more in 2019 than 2018 – and receiving ratepayer curtailment payments that are likely around a billion dollars annually. (This overproduction appears to be an example of California energy policy being set by subsidy-seekers instead of being driven my rational market design.) Furthermore, neither solar nor wind are “dispatchable.” – a major shortcoming.

    California has several large pumped hydroelectric storage systems such as Helms and Castaic. However these plants appear to not be used for bulk energy storage, based on their abysmally-low capacity factors – well below 10% per the US EIA. In fact, the average daily usage of the massive Helms Pumped Storage facility was only 12 minutes per day in 2005. More details are found here: Instead, these energy storage systems appear to be used to provide “ancillary services” that maintain voltage and frequency stability. Solar and wind destabilize the grid. Finally, Battery Electric Storage Systems are hideously expensive. They are too small by about a factor of 10,000.

    Given the poor alignment between the peak production of solar power when the Sun is directly overhead and the late-afternoon to early-evening load peak, the result is that the natural-gas-fired generation that “firms” solar and wind must be dispatched in an inefficient and intermittent fashion that likely nullifies any environmental benefits of solar and wind. This problem with solar and wind has been known for some time. Some natural gas vendors are finally admitting in advertisements that they compensate for the substantial intermittencies of solar and wind.

    For all of the above reasons (and several more) California should have spent the billions of dollars expended for short-lived solar and wind installations on dispatchable, cost-effective, reliable zero-emissions nuclear power. Nuclear power plants like DCPP are designed to last a century. The eminent scientists and engineers of the California Council on Science and Technology (CCST) were requested by the California Energy Commission (CEC) in 2010 to chart a cost-effective pathway forward for California to achieve its mandated emission targets – and facilitate electrification of the highly-emitting transportation sector. Their answer was simple. Build about 30 new nuclear power reactors in addition to the four already in operation. Fossil fuel interests made sure the pair of 2010 CCST reports were ignored. You may learn more about these controversies at the Californians for Green Nuclear Power, Inc. website at CGNP is a CPUC intervenor is several relevant proceedings, including the current R1602007 Resource Adequacy proceeding. CGNP continues to advocate on the basis of science, engineering, and economics for the continued safe operation of Diablo Canyon Power Plant (DCPP) beyond 2025. If the highly-functioning DCPP is closed, power rates will climb even higher as new dirty fossil-fired generation replaces the equivalent of five Hoover Dams of zero-emissions electricity. As another Energy Institute at Hass paper by Lucas Davis established, when San Onofre Nuclear Power Station (SONGS) was closed in January, 2012, it was replaced with fossil-fired generation.

  7. [unless i missed it as I speed read the post on my phone before getting off at BART Downtown]

    A point not mentioned is the classic semiconductor [and others] experience curve. The key is in rising volumes. I will leave out the many jokes about selling at a loss and then ‘making it up in volume’ etc. So for every doubling of cumulative volume costs [not prices] drop x %, where the x varies by industry.

    Another expression of the experience curve is Moore’s Law – chip performance doubling every 18 or so months.

    For panels, initially the silicon was a major part of cost. But as volume increased that share dropped, and the cost of ‘assembly’ of the physical panel took over. With automation of assembly that dropped too. Now the installation on-site ‘labor’ cost is probably largest factor. There are moves to integrate/ automate parts of that too. The last will be regulatory/ permitting costs.

    My concern re solar/PV is about mitigating end-of-life disposal of the toxic materials.

    [by the way it is Sharp not Sharpe – i think{

Leave a Reply to Richard McCannCancel Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: