This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.


What are the Risks Related to Solar Power Investment?

Schalk Cloete's picture
Research Scientist Independent

My work on the Energy Collective is focused on the great 21st century sustainability challenge: quadrupling the size of the global economy, while reducing CO2 emissions to zero. I seek to...

  • Member since 2018
  • 1,002 items added with 389,693 views
  • Dec 24, 2018


  • A cumulative cash flow analysis is presented for utility-scale solar PV.
  • The large effect of discount rate on levelized costs is illustrated. 
  • Value declines and integration costs cause a $60/MWh increase in the levelized cost under the base-case assumptions. 
  • This translates to a very large 13% decrease in annualized investment returns if solar market share increases by 1% per year over the plant lifetime.
  • This investment risk is not as large as it seems because the steady increases in solar market share that cause these negative returns will never happen if solar generators are not shielded from their value declines and integration costs.   


An earlier article offered some qualitative discussions on the risks involved in several mainstream energy options. Following the previous article on onshore wind, the next four articles will present a quantitative analysis of these risks for utility-scale solar PV, nuclear, natural gas and coal. The analysis will be presented for a typical developed world scenario. Developing world technology cost levels are very different and will be covered in a future article. 

All the most influential assumptions will be clearly explained and their impact on the results will be quantified. This will give the reader the opportunity to clearly see the quantified impact of the risk under the assumptions they think are the most appropriate.


Results will be presented in the form of a discounted cash flow analysis for only 1 kW utility solar PV over a one year construction period followed by an operating period that lasts for as long as the plant is profitable or up to a maximum of 40 years. The investment is made in the first year, followed by the annual receipt of revenues from electricity sales and payment of operating and maintenance (O&M) costs.

Capital costs are taken as $1800/kW. This was found to be a good global average when adjusting for purchasing power parity (see previous article). O&M costs are taken as 1% of the capital cost per year and these costs are assumed to increase linearly by 1% per year. Plant output is assumed to fall by 2% in the first year and linearly by 0.8% per year afterwards.  Inverter replacement every 15 years is included at a cost of $100/kW (hardware and installation).

After the initial $1800 capital investment, the annual cash flows from electricity sales at an average wholesale price of $60/MWh and a capacity factor of 18% are shown below. The linear decline in plant performance is clearly visible, as well as the linear increase in O&M (although O&M costs are relatively small). Inverter replacement costs are also notable. Note that the 18% capacity factor is selected to be optimistic next to the global average that has hovered around 15% for the past 4 years according to BP data.

Using this information, a cumulative cash flow curve can be constructed (below). As shown, the initial $1800 investment is recovered in year 28 when no discounting is applied (discount rate of 0%). When a discount rate of 1.4% is applied, the net return on investment is zero. In other words, this analysis would return a levelized cost of electricity of $60/MWh if the discount rate is set to 1.4%. Under a more realistic discount rate of 8%, the initial investment cannot be recovered.

Subsequently, the effects of the value declines and cost increases related to intermittency (discussed in the previous article) are included. Firstly, the added cost of grid connection is included as an up-front cost. It is assumed that the average distance between the solar farm and the consumer is 100 km, yielding an added capital cost of $200/kW at a transmission cost of $2/kW/km.

Secondly, balancing costs are assumed to scale directly with the wind energy market share, adding $0.3/MWh for every percentage point of market share. This is about half the current balancing cost in Germany.

Thirdly, the value decline of solar power is modelled according to the following market value factor. At market shares higher than 15%, the linear trend is extrapolated. It should be mentioned that this trend is representative for Europe, China and Japan (representing 74% of current installed capacity) where solar capacity factors are quite low. Markets like the US South-West, Australia and the Middle East will see less pronounced value declines resulting from higher capacity factors and a better match with seasonal demand. 

Wind and solar value factors (a value factor of 1 is for a generator with a constant output) as a function of their respective market shares (source).

The following annual cash flows are generated when these assumptions are applied to a plant constructed when the solar market share is 2% (current global average) and increasing by 1% per year (up to a maximum of 40%). The more rapid decline in revenue (caused by the value decline) and the increased balancing costs are clearly visible.

As shown in the cumulative cash flow analysis below, not even half of the initial investment can now be recovered, even under a 0% discount rate. The plant starts making a loss at the time of the first inverter replacement at year 15 as declining revenues fall below increasing costs.

Effect of the discount rate

The effect of discount rate on the average electricity price required is shown below with and without the value declines and cost increases from intermittency. Note that the average electricity price required is used here instead of the levelized cost of electricity to account for the value decline of solar power with increasing market share. This measure can be interpreted as the average market price over an entire year that will yield a zero return on investment with a specified discount rate. The actual electricity price received by the solar farm will be lower.

The graph shows that the required electricity price almost quadruples as the discount rate is increased from 0% to 15%. Inclusion of the value decline, balancing costs and grid costs increases the required electricity price by about $60/MWh at 0% discount rate with a moderate increasing trend towards higher discount rates. This impact is double that of wind power because of solar's more pronounced intermittency and high correlation with other solar generators.

Quantifying the risk

Next, the influence of the risk of accountability for value declines and cost increases caused by intermittency will be quantified. This quantification is done by determining the discount rate giving zero return on investment when the average electricity price is set to $60/MWh. The annualized return on investment is then quantified as the discount rate minus 2% to account for margin erosion from technological improvements of new plants that come online during the plant lifetime as well as financial/legislative costs (paying the bankers and lawyers involved in setting up financing for the plant).

As shown below, the investment return is a little less than 0% when solar farms carry no accountability for intermittency costs (blue bar). The sensitivity to the three different intermittency effects is shown by the orange bars. When a 100 km grid connection is included, the annualized investment return drops by only 0.6%. A much larger 8% drop occurs when the value decline is added. Further addition of balancing costs also has a large effect, reducing the annualized investment return to -13.5%.

The magnitude of the drop in investment returns is strongly influenced by the rate of solar power expansion over the lifetime of the plant (grey bars). More solar on the grid will reduce the value and increase the balancing costs of all solar generators. An increase in the rate of solar expansion from 0.5% to 1% per year lowers the investment return by 7%, while a further increase in expansion rate to 1.5% per year cuts another 5% off the annualized return.

The effect of added grid expansion costs (yellow bars) is smaller. Increasing the required grid connection from 100 km to 500 km (thus increasing the added up-front cost from $200/kW to $1000/kW) only decreases the annualized investment return by a little over 2%.

Finally, the effect of balancing costs is shown by the green bars. Every increase of $0.2/MWh per % of wind market share decreases the investment return by a little over 2.5%.


This article has quantified the large negative effects of solar intermittency on project economics. The potential that these costs will eventually be fairly attributed to solar generators presents an important risk for solar farm investors.

Even when these intermittency effects are completely ignored, the global average solar farm does not give any return on investment without direct subsidization. As always, however, I have to stress that there are locations where solar is much more attractive. For example, the US South-West can achieve very impressive capacity factors up to 30% thanks to an excellent solar resource and high inverter loading ratios (which also increases capital costs to about $2200). Under these assumptions, the annualized investment return amounts to 2.2% (with no accountability for intermittency effects). Such high capacity factors will also significantly reduce the value decline. 

When intermittency costs are correctly accounted for, the investment returns fall drastically - substantially more than for wind power. The annualized return on investment drops from -0.6% to -13.5% when intermittency costs and value declines are accounted for under the base-case assumptions. This very large decline can be moderated by battery storage systems because of solar's regular daily production cycle. A future article will explore this potential mitigating effect. 

The most influential factor in the analysis is the rate of solar power expansion. Higher expansion rates lead to larger investment losses. Interestingly, this dynamic actually reduces the investment risk because the steady increases in solar market share that cause this risk will not take place if value declines and integration costs are fairly assigned to solar generators.

This reinforces the earlier notion that continued solar power expansion will require perpetual subsidies. If solar power is to have the impact envisioned by green advocates, the large investment losses caused by their value declines and integration costs will need to be borne by other sectors of the economy for decades to come.

Schalk Cloete's picture
Thank Schalk for the Post!
Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.
More posts from this member
Spell checking: Press the CTRL or COMMAND key then click on the underlined misspelled word.
Joe Deely's picture
Joe Deely on Dec 25, 2018

Let's compare a couple of assumptions in this article to other reported US numbers.

1) Above article used the following for solar capital costs:

Capital costs are taken as $1800/kW.  (In other words $1.80W)

SEIA reports:

Utility scale fixed-tilt at $0.93/W and single-axis tracking at $1.04/W for 2017Q3. Note the "despite tariffs" reference.

Prices across market segments are now all at historic lows despite tariffs on modules, inverters, aluminum and steel: $2.85/Wdc, $1.47/Wdc, $0.93/Wdc and $1.04/Wdc for residential, non-residential, utility fixed-tilt and utility single-axis tracking systems, respectively. 

In other words even with tariffs - costs have dropped to all-time lows.

Reported costs are 44% lower than assumed costs.

2) Above article used the following assumption for Capacity Factor:

 Note that the 18% capacity factor is selected to be optimistic next to the global average

Actual reported average US capacity factor from EIA - 2017.


 In the US, average capacity factors range from 18% in MN to 33% in AZ.

The assumed CF number used is 44% lower than the actual average.

3) Finally, let's look at what solar investors in the US are actually building currently and the prices they are getting from utilities.

Utilities across the US are signing solar PPAs for less than $30/MWH.

      New Braunfels in TX - 255MW 15 yr PPA @ $25/MWh

      NV Energy 6 PPAs below $30/MWh

many more...

Next up solar and storage.



Schalk Cloete's picture
Schalk Cloete on Dec 27, 2018

Hi Joe, there are some simple explanations for the points you raised. 

Firstly, the capacity factor you quote is for the plant capacity defined in terms of AC power output. In general, US solar farms feature an inverter loading ratio of a little over 1.3, so your cost numbers (per kW-DC) should be multiplied by about 1.3 to match to the capacity factor you quote. 

Secondly, there are several neglected cost components in typical low solar PV cost estimates. The Lawrence Berkely utility PV study comparing empirical PV price data to modelled estimates lists the following potential omissions: "For example, GTM represents only turnkey EPC costs and excludes permitting, interconnection, and transmission costs, as well as developer overhead, fees, and profit margins." 

It is based on these empirical cost estimates that I calculated the expected investment return of 2.2% (with no intermittency costs) for the US case in the conclusions of this article. 

The impressively low PPAs is simply due to the subsidies, mainly the ITC. 

That being said, as mentioned twice in the article, solar resources in places like the South-West US is much better than the global average, reducing both project costs and intermittency costs. This article is meant as a global average indication of solar PV economics and it is clearly acknowledged that the US will be significantly better than average. 

Joe Deely's picture
Joe Deely on Dec 28, 2018


Thanks for the correction on Capacity Factor. Will keep that in mind.

As for your other points:

"For example, GTM represents only turnkey EPC costs and excludes permitting, interconnection, and transmission costs, as well as developer overhead, fees, and profit margins." 

Actually the SEIA report I referenced does include permitting, overhead and profit margin. See legend for chart.

Note also that my link was from Q3 of 2018 vs the 2017 LBNL survey you used and therefore includes 2018 price declines.

That said  - I am fine with your source... 

However, if you are starting a solar project now then you need to look at what prices will be in 2019/2020.  This is when you would be buying most tof the modules.

Looking at Fig 8 from your LBNL 2017 survey of solar projects - 8% of the projects had costs betwen $0.75 - $1.25 in 2017. Costs have continued to fall - so that share will be higher in 2018. What will it be in 2020? Let's go with 50%.

Note also the possibility of dropped solar and steel tariffs by 2020.

Couple of other things.

The solar in Nevada - as in much of the West outside of CA - is replacing coal not NG. So you would need to use coal CO2 numbers in your calculation for CO2 avoidance.

Finally, some of the NV Energy PPAs include storage. This changes the "value" equation significantly.  Going forward, this will be common. NREL just came out with their first Solar+storage benchmark.

Pretty good prices. But again as an investor, I would be wondering what these prices will be in 2019/2020 when the material for my project is being sourced?

Schalk Cloete's picture
Schalk Cloete on Dec 28, 2018

Yes, the SEIA costs are impressively low. It is worthwhile to note, however, that SEIA's projections (when using these low installation costs) show a gradually declining utility solar installation rate of about 8 GW-DC/year by 2023 when the last plants qualifying for the ITC will have to be constructed. There will probably be a steeper decline from 2024 onwards as is projected in the nearer term for wind power following the PTC phase-out. 

An 8 GW/year installation rate with a 30 year operational lifetime and a capacity factor of 18% (accounting that this is DC capacity and will degrade over time) amount to a steady-state electricity production of 378 TWh/year - about 9% of total electricity (4% of primary energy) for one of the better solar countries on Earth. 

About the addition of storage, it is good that solar is already being asked to mitigate its intermittency by incurring additional costs. For the example you show, the large-scale co-located solar+storage system costs about $1870/kW-DC and will probably operate at an effective capacity factor of below 18% (accounting for cycling losses). It will also need a couple of battery replacements over the lifetime of the PV plant. These numbers will result in decidedly negative ROIs even in the US with its excellent solar resource: much worse in Europe, China and Japan where three-quarters of current global capacity is installed. 

Joe Deely's picture
Joe Deely on Dec 28, 2018

Yes, the SEIA costs are impressively low.

Actually, I'm not really impressed. All I care about is that costs continue to decline. Amazing that costs declined in 2018 - even with solar and steel tariffs. Let's see how low costs get by 2023 - especially with removed tariffs. Then maybe I'll be impressed.

it is good that solar is already being asked to mitigate its intermittency

Xcel RFP returned median bids with solar plus storage at $36 MWh. Wind with storage at $21/MWh.  Goodbye coal in Colorado.

For the example you show, the large-scale co-located solar+storage system costs about $1870/kW-DC

I'm not that sure that all of the intial storage projects will be co-located. For example, in southern CA many of the existing solar projects are out in desert. The storage will probably be placed within urban LA - where it has much higher grid value.

There are a bunch of NG plants closing down in LA area over the next few years - these would be great spots for large storage projects.

Bruce McFarling's picture
Bruce McFarling on Dec 29, 2018

It appears that what this is saying:

That being said, as mentioned twice in the article, solar resources in places like the South-West US is much better than the global average, reducing both project costs and intermittency costs.

... is that this article is using "global average" with reference to the current global installed base, heavily biased by the particular history of policy regimes toward a region of the world with substantially less solar potential than the majority of the world's population.

Finding out that the case for solar power is not very strong when it is based in a large part on a part of the world close enough to the North Pole poles that it only has the population it does due to an accident of warm ocean currents ...

... is a bit less broadly applicable than the "global average" would suggest at first glance.

Schalk Cloete's picture
Schalk Cloete on Jan 2, 2019

This is a fair point. I could only find high-quality data on the value decline and integration costs for Europe and tried my best to clearly state the impact of this assumption in the article. 

That being said, China will remain the major growth centre for solar PV over coming years. Chinese solar resource quality is very poorly correlated with population density, giving it an average fleet-wide capacity factor as low as Europe (around 13% according to BP Statistical Review data). 

India is seen as an important solar growth market in the longer term, but the Indian solar resource is not that great either (about 20% capacity factor). The real solar quality is found in the Middle East (where it competes with $10/barrel oil), Africa (where it will be decades before the economy reaches a meaningful size on a global scale), Australia (which only houses 0.3% of the global population), and the South-West US and Mexico (probably the best growth opportunity for high-quality solar). 

I therefore think this is not such a bad reflection of the global average situation for the foreseeable future, although it is a bit slanted to the pessimistic side (as clearly acknowledged).

Bruce McFarling's picture
Bruce McFarling on Jan 5, 2019

But in India, that is competing against more expensive electricity, which "globalizing" the value of the power ignores.

Here in China, with the limited domestic NG resource and the very high full economic operating cost of coal when CO2 emission costs are taken into account, the high quality Tibetan solar resource can certainly justify HVDC to Hubei & the Three Gorges Dam interconnects. The Southeast where the largest concentration of the population lives tends to be fairly rainy ... and after all, it was that plus the long growing season allowing two rice crops per year that originally explained why such a large share of the population lived there. And those population concentrations bias the CF of the installed solar base.

Which highlights the status quo bias of using the CF of existing installations in a country to predict CF of future resource opportunities. 

What you'd primarily place in the Southeast would be decentralized solar PV and thermal (heating & hot water), while more of the bigger concentrated solar PV as well as CSP would be placed in the drier southwest where there is a much better solar resource ... and where the time of generation complements the drop off in generation of distributed solar in the Southeast.

Wayne Lusvardi's picture
Wayne Lusvardi on Dec 26, 2018


What would be the ROR generated in your cash flow model by using Joe Deely's assumptions shown above?  Would the discount risk rate also have to be adjusted also under Deely's assumptions, not just mechanically modifying the cash flow? 

Is the PPA contracts for $25/MWh in New Braunfels TX and $30/MWh in Nevada an apples to apples comparison when it doesn't appear to account for subsidies/cost shifting, balancing costs, extra connection costs, etc. in your model?  

What would be the estimated cost per ton of carbon removed, also considering the California Cap and Trade tax on electricity at the user end of the chain? 



Joe Deely's picture
Joe Deely on Dec 26, 2018

Here is NV Energy IRP - if you want more information on these solar projects and their "connection" costs. (Pg 6)

Added transmission will cost  $20.7M  which is less than $0.50 per MWh over 20 years.  A blip.

Nevada(NV) is closing its last utility coal plant AND eliminating its share of Navajo coal in AZ - 2019. Much cleaner grid.

Hello solar/storage. Goodbye coal.



Schalk Cloete's picture
Schalk Cloete on Dec 27, 2018

The ROR for an example case from the US is estimated at 2.2% in the conclusions of the article based on recent empirical cost data. 

For very high capacity factor plants like those that should be possible in Texas and Nevada, the negative effects of accountability for intermittency costs in the final figure in the article should be smaller, but even if we assume that these investment losses are halved, they are still large. 

Estimating CO2 avoidance costs is challenging because there are many assumptions involved. However, if we ignore value declines and integration costs and assume that solar displaces natural gas at 400 g-CO2/kWh, the 30% ITC pays solar developers about $600/kW-AC, which amounts to about $28/MWh at an 8% discount rate and 30% capacity factor. The ITC therefore pays about $28 for every 0.4 tons of CO2 avoided or $70/ton. I estimate that value declines and integration costs will roughly double this cost over the plant lifetime. 

Wayne Lusvardi's picture
Wayne Lusvardi on Dec 27, 2018


I assume your DCF model is for production of wholesale power for re-sale to IOU's, in which case the PUC-approved ROR would have to be added to develop the retail price.  Southern California Edison, San Diego Gas and Electric and Pacific Gas and Electric approved ROR for 2018 is 7.61%.  If the retail price of electricity is what is desired to be estimated then the ROR must also be added. 

I value California PU-regulated private water utilities and there is a market relationship between net income produced per Kwh or meter and total value. 


Jim Stack's picture
Jim Stack on Dec 28, 2018

I don't want to be to simple but Nuclear gets the biggest subsidies of any power, we also still subsidies NG, COAL . Yet you state that Solar needs subsidies QUOTE=Even when these intermittency effects are completely ignored, the global average solar farm does not give any return on investment without direct subsidization.

Wayne Lusvardi's picture
Wayne Lusvardi on Dec 31, 2018

It depends on what you mean by subsidy. Does that mean tax depreciation?  Every industry gets that on plant and equipment.

And you may have missed the point of the article. If concentrated solar power plants have negative to miniscule rates of return then government in one way or another is subsidizing the entire installation and operation by mandating a market to buy solar power. This is one step away from nationalizing an industry as is done in socialist and communist systems. Unless I misunderstand, solar power companies are making a ROR based on nearly pure subsidy. Regulated IOU's in California are allowed a 7.61% ROR on the weighted average of cost of capital/equity.  But unregulated merchant solar power companies are making a return on subsidies. Under California's constitution, that might be considered an illegal gift of public funds except that public policy dictates such subsidies. 

And all for what? According to the US EIA and the Haas Business School at UC Berkelty (not the California ARB), California has reduced C02 from about 425 million metric tons in 1990 to +/- 425 MMT in 2016; or negligible. That is after Cap and Trade kicked in around 2012 and 30% of electricity was mandated to renewable energy. And that does not include the 68 M/T of C02 from 2018 California fires.   

Bruce McFarling's picture
Bruce McFarling on Jan 1, 2019

The biggest subsidy is, of course, permission to use the atmosphere as a CO2 dump for free or at steeply discounted prices, but, yes, special tax treatment for fossil fuel producers ~ allowing percentage depletion of oil reserves rather than cost depletion, as if it was a depreciating real capital asset, not requiring intangible drilling costs to be depreciated but instead allowing immediate expensing, and allowing natural gas production to use the manufacturing deduction ~ are all additional subsidies to natural gas fueled electricity production.

Profit is a residual ... an additional $20-$45/MWh for NGCC at realistic carbon prices and substantially more for coal fired power would change the IRR of solar substantially.

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »