Two years ago we published a piece as part of a three-part series analyzing CSI data in SCE’s territory where we operate. To this day it remains one of our most popular posts ever, with over 11,000 views. A year later we repeated the analysis with that year’s data - and that post was also very popular, easily becoming our most viewed post of the year. In both of those posts, SolarCity - and its curious practices with long delays in building systems and unusual system pricing - was featured prominently. We didn’t plan it that way - indeed the title of the posts, Outliers and Oddities, reflects how these were stories that simply fell out of the data.
Still, as any blogger can tell you, blogging can be a lonely business and it is often hard to know that anyone is really paying attention.
But yesterday changed that.
Yesterday we were informed of a report about SolarCity done by the firm Copperfield Research appearing on the investment site, Seeking Alpha. That 30-page report, titled SolarCity (SCTY) - The Emperor(’s cousins) Have No Clothes - uses multiple charts and graphes from our analyses to support their conclusion that SolarCity is seriously overvalued in the market and that “The bottom line is investors have been hypnotically drawn to SolarCity, like moths to a flame, dangerously ignoring the house of cards on which the story appear to be built.” Ouch.
Indeed, the report has three full pages devoted to our analyses - here’s a sampling of their take:
Run on Sun Founder and CEO Jim Jenal performed a rigorous third party study examining many of the solar installers and developers. IT SHOULD BE MANDATORY READING FOR ANY SOLARCITY SHAREHOLDER OR DEFENSIVE ANALYST. One emphasis for Jenal’s study was his concern that solar developers had been violating a critical segment of the Solar Energy Industry Association’s Solar Bill of Rights. Bullet #8 of the Bill of Rights states, “Americans have the right to - and should expect - the highest ethical treatment from the solar industry.” Jenal believed that there were certain “situations that don’t live up to that Right.” His exposé spent significant time scrutinizing SolarCity’s cost metrics…
Jenal’s comprehensive analysis concludes that this prospective over-reporting of FMV [fair market value] appears to have been occurring consistently dating back to 2008, with the SolarCity tax fraud [their words, not ours] (if FMV was misrepresented) potentially costing tax payers many-millions of dollars!!!
But alas, again due to the sensational analysis by Jenal…
(Emphasis in the original.)
And on it goes (you get the picture).
As we noted at the outset, we didn’t set out to write an “exposé” of SolarCity. Rather, we were simply trying to understand the market in which we operate and in the course of doing so we stumbled upon the oddities that we reported. But it is certainly gratifying to see that others have found value in what we published.
Now if we could only get someone to pay attention to what is going on in Glendale…
Technology reporter Felicity Carus of the (Manchester, UK) Guardian newspaper has a new article up on PVTech that features an extended interview with Run on Sun Founder & CEO, Jim Jenal:
For all your CSI data crunching needs, let me introduce readers to Jim Jenal, the founder and Chief Executive of Run on Sun, a small Pasadena-based solar installer … Jenal was a litigation attorney for 13 years after graduating with a BA in mathematics and a masters in computer science. He may not practise the law any longer, but his inner data geek is alive and kicking, particularly when it comes to CSI data, which he has analysed extensively in his blog.
If anyone can point me in the direction of equivalent data analysis for the CSI or other programmes in California’s solar industry, I would love to hear from you.
Carus went on to discuss our results from the Outliers & Oddities posts (which looked at some of the less savory activities in the solar industry) and ended with this quote:
Jenal thinks those in the solar industry should be good citizens. “I actually think that we’re supposed to be different. I understand those who maximise their profits. I understand that it’s much of the way the world works. But I don’t think the solar industry, which is about sustainability, should operate that way. It offends me and my sense of justice.
We would also note that it violates not just our Founder’s sense of justice, but also the Solar Bill of Rights, promulgated by the Solar Energy Industry Association (SEIA), which declares: “ Americans have the right, and should expect, the highest ethical treatment from the solar industry.“
We are thankful, on this Thanksgiving eve, for the overwhelming majority of honest, ethical players in the solar industry who collectively do make us “different". But we still have an obligation to not only behave ethically ourselves, but to call out those few who aren’t.
Happy Thanksgiving, everyone!
UPDATE x2 11/8 - Solar City’s Jonathan Bass adds his perspective on our reporting about Solar City - see his response in the comments.
UPDATE 9/30 - We just heard from Jonathan Bass at SolarCity. Details at the end.
(Still no word from Galkos!)
Editor’s Note: We have now done an updated analysis showing the same data from 2012. You can read our 2012 Outliers & Oddities here.
In the first two installments in this series (Part 1 and Part 2) we looked at the most recent data from the California Solar Initiative (CSI) covering the first half of 2011 in SCE’s service area. Using that data we identified trends in cost, equipment and system efficiency. Along the way, we stumbled upon some Outliers and Oddities in the data that left us puzzled and disturbed. In this post we name names, specifically Galkos Construction (aka GCI Energy) and SolarCity.
Before we explain to you why they are featured in this post, we would remind our readers of the Solar Bill of Rights created by the Solar Energy Industry Association (SEIA) in the Fall of 2009. We wrote at some length about the Bill of Rights when it was introduced, but we want to highlight now what then we termed to be, “the most important right of all:”
8. Americans have the right, and should expect, the highest ethical treatment from the solar industry.
Beyond a shadow of a doubt, this is the most important Solar Right of all if we are to build an industry that is respected and trusted by consumers throughout this country. This should almost go without saying - and yet, saying it, and living it, is extremely important.
In our view, if we become aware of situations that don’t live up to that Right, we have an obligation to point them out so that our potential clients can make the most informed decisions possible.
In honor of that principle we present today’s post.
In looking at the data, from time-to-time a data point would jump right off the screen. For example, examining all of the residential projects in our data - both “completed” and “pending” but excluding “delisted” - we find that the average installation cost in CSI Rating AC Watts is $8.43/Watt (in DC or nameplate Watts that average becomes $6.99). As we noted in Part 1, that number has decreased over time and also decreases as system size increases. Still, given that the residential sector (as designated in the CSI data) only consists of systems between 1 and 10 kW, you wouldn’t really expect significant price variation between installers over a six month period.
But you would be wrong.
Here is a chart of the Cost per Watt for the largest installation companies in the SCE service area (you can click on the chart to see it full size):
First, let us give credit where it is due. The low end outlier is HelioPower, Inc., at $6.56/Watt, and they did it with an efficiency factor of 87% - second best of anyone on that chart. Nice.
But who is that way off in left field? Coming in at a staggering $13.32/Watt - a full $1.40 higher than their nearest competitor and more than twice what HelioPower is charging - is Galkos Construction, Inc., also known as GCI Energy, out of Huntington Beach. For that money, they must surely be offering only the most efficient and sophisticated technology, right? Not so much. To the contrary, the average installation efficiency for Galkos is only 84.9% - the second worst on the chart and well below the average of 86.11%. In fact, 99% of the time Galkos appears to use Sharp panels - not exactly an exotic solar panel brand - and in particular the Sharp ND-224UC1 panel (66.5%). A quick Google search reveals that the Sharp ND-224UC1 can be purchased, at retail, for $2.65/Watt or less. Given that Galkos handled 400 projects in this data set, it is hard to believe that their price for all of their equipment, particularly the Sharp panels, would not be heavily discounted.
Quality, of course, is important, and the data does not reveal - though the Internet hints at - the quality of installations from Galkos. Here is how the company describes its own product offerings (from the “Services” page of their website):
Solar by GCI [Galkos Construction, Inc.] Energy
GCI Energy is the largest solar company in Southern California with over 30,000 customers. So you get the most knowledgeable professionals, excellent customer service and a better price.
GCI Energy solar offers the highest efficiency solar panels on the market - those manufactured by Sharp. With Sharp Solar Panels, GCI Energy can tailor a solar panel installation to your specific needs and lifestyle, so you get maximum performance without a maximum investment.
Does Galkos actually have 30,000 solar customers? Certainly not (nobody does). Are they providing “a better price"? It is not clear what their standard of comparison might be - but their price is not better than any of their major competitors in that chart. And of course, the statement does not define what they mean by “the highest efficiency solar panels on the market,” but it seems unlikely that Sharp would make that claim. Here’s one chart that concludes that they couldn’t (note the efficiency of the SunPower and Sanyo panels first, then search for Sharp).
All we can say in response is, caveat emptor.
Now we turn to the Oddities section of this post. Unlike the outliers, which were always of interest to us, we were not looking for the oddity we report here - it literally just jumped out at us.
Question: What is the difference in reported cost between systems sold directly to the end customer and those that are leased (i.e., have a third-party owner in CSI parlance)?
The initial difference that we stumbled upon was so startling that we knew we needed to narrow our focus and control for as many variables as possible to isolate that one factor. To achieve that end we restricted the data to those residential systems (i.e., between 1 and 10 kW) that were “pending” in the CSI/SCE data (thus, the newest proposed systems in the data which, based on our Part 1 analysis should mean the lowest cost systems). That way our project sample would be as homogenous as possible, eliminating cost variations based on system size and timing.
Given those restrictions, the top 5 installation companies in which the system is owned by a third party are: Verengo (482 systems), SolarCity (468), American Solar Direct (124), Sungevity (99), and HelioPower (63). Of those five, only two also have direct sales projects pending: Verengo (7) and SolarCity (9). Let’s see how they compare:
What is going on here? For Verengo, as the number of systems increases - which it does in going from sold systems to leased systems - their cost per Watt decreases - which is what we would expect. But not so for SolarCity - even though they are leasing 50 times as many systems as they are selling, their cost for the leased systems went up - way up - as in up by $3.12/Watt!
(One possible explanation for this discrepancy would be that SolarCity uses much more expensive equipment in their leased systems than they do in the ones that are sold. But they don’t. On their sold systems, SolarCity always selected a Fronius inverter and their panel choices were split among Yingli (56%), Kyocera (33%) and Sharp (11%). On their leased systems, SolarCity selected Fronius inverters 98% of the time and again split their panel choices among Yingli (68%), Kyocera (28%), and BP (3%) with the remaining 1% scattered among Suntech, Sharp and Sanyo. In other words, there is no significant difference in SolarCity’s equipment choices between sold and leased systems.)
Why does this significant cost differential matter, you might ask? After all, customers aren’t paying that price - they are paying on a lease so the “cost” of the system doesn’t matter to them, all they care about are their lease payments. True enough - unlike the case with our Outlier above, the end customer is not the victim here.
Recall, however, that for systems that are leased, the third-party owner - presumably SolarCity and its investors in this case - receives both the rebates and the tax benefits associated with the installation. While the rebates are independent of the system cost (they are paid based on CSI Watts), not so for the tax benefits. Commercial operators (even though these are residential installations they are treated as commercial projects for tax purposes) are entitled to both a 30% tax credit as well as accelerated depreciation based on the cost of the system.
For the 468 systems that SolarCity is leasing, their total cost is $24,261,735 to install 2,412 kW. If those installations were billed out at the $6.94/Watt they are charging for their sold systems, the installed cost would be $16,739,280 - a difference of $7,524,037. At 30% for the federal tax credit, taxpayers are giving SolarCity an extra $2,257,211 - just from six months worth of installs in only the SCE service area.
In the words of the 70’s pop song, How long has this been going on?
We decided to find out.
Although all of our analyses up until now in this series have been restricted to the first half of 2011, the actual data set contains entries from the inception of the CSI program. Thus we can look at all of SolarCity’s installs going back to 2007 and compare them as we did for the 1H2011 pending installs above. We will use the First Completed date to group these by year and analyze only “installed” - and not “pending” applications. Here’s the data:
The answer would appear to be, almost from the beginning! Back in 2007, Solar city sold ten times as many systems as it leased. By 2008 the ratio was down to 4-1 and ever since then leasing has been SolarCity’s predominant business strategy with the ratio of leased to sold now standing at nearly 16-1 in 2011.
What, then, is the cumulative impact to SolarCity’s bottom line from this trend throughout California? We aren’t in a position to calculate the depreciation benefits (since that is a function of the system owner’s tax bracket) but we can readily calculate the added value derived from the 30% federal tax credit due to this increased cost per Watt.
Here is our plot of the cumulative effect of those year-by-year increases:
After a slow start in 2007-08, SolarCity’s “model” really took off and has garnered the company an extra $3,000,000+ each year since 2009 (and, of course, 2011 is not yet over) for a total excess accumulation of $10,619,000. Depending on the investors’ tax bracket, the depreciation could be worth nearly as much as the tax credit.
We just heard from Jonathan Bass, Director of Communications at SolarCity who took exception with our report, although he did concede that he could see how we could have reached the conclusions we published in light of the CSI data. We encouraged him to please send us a written response in as much detail as he chose and we would publish it in its entirety. While he agreed that SolarCity would be publishing its response, he did not commit to publishing the information here.
In any event, when we hear more we will update this post again.
No doubt there is more that we could do with these revelations - but wouldn’t it be better for those with actual oversight obligations to examine this data as closely as we have and to take appropriate action?
As always, we welcome your comments - and if we hear from any of the folks named in this series we will be sure to update the appropriate post.
Yesterday we wrote about our most recent foray into the California Solar Initiative (CSI) data and how that data revealed trends regarding the costs of solar in SCE’s service area during the first half of 2011. We continue today with a look into the equipment that was specified for these projects and explore who’s hot and who’s not.
As a reminder, our data set for this analysis consists of an extract from the CSI Working Data from 8-24-2011 that includes data for SCE installs where major status activity took place during the first half of 2011. That data set consists of a total of 6,306 projects of which 698 are “delisted” (meaning the project’s rebate reservation has been cancelled for some reason), 3,131 are “installed” (completed or in some stage of rebate payment) and 2,477 are “pending” (in some stage of the process from initial rebate application filed but no rebate claim yet filed). For today’s analysis, we will exclude the “delisted” projects from our data, leaving a total of 5,608 projects to analyze.
CSI tracks data about equipment used on projects in great detail. In particular, for every project, CSI allows for up to seven different panel manufacturers and ten different inverter manufacturers! So how many of our projects use multiple panels or inverters by different manufacturers? We would expect not many, and the data supports that surmise. Only 11 projects reflect two different solar panel manufacturers on the same project and in most of those cases the installer has substituted one solar panel for the one originally designated. (Indeed, one project reflects four different panels being identified to CSI for the same project, finally settling on what appears to be two 180 Watt panels plus six 175 Watt panels feeding a single string inverter - curious design, that!) Similarly for inverters, only 24 projects have two different inverter manufacturers specified and no project reflects more than two. Given that, our analysis will only look at the first specified panel and inverter manufacturer.
So what is happening with solar panels?
Overall, there are 85 different panel manufacturers included in the data; however, most of them account for very few projects. If we apply a reasonable filter to this data and only look at solar panels that appear in 50 or more projects, the number of represented manufacturers drops from 85 to 14, and the total number of represented projects falls from 5,608 to 5,079. In other words, those fourteen manufacturers account for 90.6% of all of our projects, as demonstrated in this first graph. In fact, the distribution is even tighter with only five manufacturers exceeding 10% of the total: Suntech Power (18.7%), Sharp (14.8%), SunPower (14.3%), Kyocera (12.8%) and Yingli Green Energy (10.8%).
There are twenty-three different inverter manufacturers represented in the CSI data, reflecting the greater complexity of inverters and the more rigorous path required to bring an inverter to market in the U.S. Filtering for manufacturers represented by ten or more systems cuts the list from 23 down to just 13.
SMA America is the runaway winner in this competition. Under their own label, they account for 35.4% of all of these projects. Moreover, the majority, if not all, of the “SunPower” inverters are actually re-branded SMA inverters. When the SunPower inverters are added in, SMA accounts for a whopping 48.2%. That leaves only six other manufacturers to exceed even 1% of the total: Fronius USA (19.8%), Enphase Energy (15.2%), PV Powered (5.5%), Kaco New Energy (3.1%), Power-One (3%) and SatCon Technology (3%).
Two things of interest in those last numbers - the inroads of relative newcomer Enphase Energy (which was only founded in 2006), and the inclusion of SatCon, since alone among that list, it only sells central inverters for the commercial market (where it is dominant).
That is how the different manufacturers stack up head-to-head, but what about combinations? Are there pairings of panels and inverters that are most commonly preferred? The data reveals five combinations that account for more than 5% of the total: Suntech panels with SMA inverters (13.5%); SunPower panels with “SunPower” (i.e., SMA) inverters (12.4%); Sharp panels with Enphase micro-inverters (10.3%); Yingli panels with Fronius inverters (8.2%); and Kyocera panels with Fronius inverters (8.2%).
While certain pairings are popular - are they cost-effective and how well do they perform together? We decided to look at system combinations from an average $/Watt perspective and from an average CEC efficiency perspective to see what jumps out of the data.
Here’s the first thing that struck us - picking a system with lower-tier panels does not guarantee a lower installation cost. In fact, many of the bottom-tier panels (none of which made the cut in our discussion of panel manufacturers above) had install costs well above our overall average for the data set ($6.37/Watt). For example, we found a handful of systems using solar panels from such luminaries as Apollo Solar Energy, SET-Solar, and REC ScanModule where the average installation cost was more than $10/Watt!
Who was on the very low end of the install cost curve? Gloria Solar, Suniva, Kaneka, Silray and Solaria each had a handful of installations that were below $4.50/Watt.
More significantly, how did our most popular pairings perform? Here’s the data:
|Combination||Average Cost $/W|
|Suntech & SMA||$5.01|
|SunPower & SunPower||$8.49|
|Sharp & Enphase||$11.62|
|Yingli & Fronius||$9.58|
|Kyocera & Fronius||$9.79|
What is up with the Sharp & Enphase combination? While Enphase installations are known to cost a bit more than a comparable string inverter installation (confirmed by our own experience), they certainly don’t cost $5/Watt more! Rather, it turns out that the overall average for all Sharp-based systems is $8.53/Watt (nearly $2.00/Watt above the average) with prices ranging from a low of $6.17/Watt (when paired with a Solectria inverter) to a breath-taking high of - are you sitting down? - $19.30/Watt when paired with a Sharp inverter. So who installed that system, you ask? You’ll learn all about it (or at least all that we can tease out of the data) later in this series.
Shifting our attention to efficiency, thin-film module maker First Solar gets the highest overall ranking, 91.7%, thanks to its extremely high STC to PTC ratio. On the more embarrassing end of the scale, Sunlan solar brings up the rear, averaging only 80.7%. From our list of the most popular solar panels, Sanyo (long a Run on Sun favorite) does the best, averaging 89.3% across a variety of inverter combinations. The rest of the top five are: Canadian Solar (87.6%), SunPower (87.5%), Suntech (87.2%), and Schuco (87.0%). The bottom-five of our best selling panels? That dubious honor belongs to: Sharp (86.0%), BP Solar (85.8%), ET Solar Industry (85.6%), Trina Solar (85.5%), and REC Solar (85.1%).
As for our five most popular pairings, here is the data:
|Combination||Average System Efficiency
|Suntech & SMA||87.2%|
|SunPower & SunPower||87.1%|
|Sharp & Enphase||85.2%|
|Yingli & Fronius||85.6%|
|Kyocera & Fronius||86.0%|
That is a pretty tight grouping, with a total range of just 2%. To break out of that mold with a conventional panel/inverter pairing, the Sanyo & SMA combination is your best bet, weighing in at 89.5%.
Finally, we decided to see what equipment combinations are preferred by the biggest installers in the market. The following table lists the top-five installers and reports the number of projects in the data, their most frequently chosen solar panel (and % of times used) and their most frequently chosen inverter (and % of times used).
|Name||# of Projects||Panel Mfr (%)||Inverter Mfr (%)|
|Solar City||910||Yingli (47.8%)||Fronius (95.5%)|
|Verengo||688||Suntech (91.7%)||SMA (81.7%)|
|Galkos Construction||401||Sharp (98.5%)||Enphase (99.0%)|
|REC Solar||207||Kyocera (42.5%)||SMA (75.4%)|
|Real Goods Solar||165||Kyocera (54.6%)||SMA (65.5%)|
Collectively, these 5 installation companies accounted for 42.8% of the projects in the CSI data. Certainly companies this large must have some real clout when it comes to negotiating prices, thereby allowing them to pass along those savings to their many customers.
Or do they?
Find out in our next installment!
The California Solar Initiative (CSI) is responsible for overseeing solar rebates for California’s three Investor Owned Utilities (IOUs): PG&E, SCE and SDG&E, and in that role the CSI program collects some very interesting data. As we have in the past, we decided to dip into the data from the first half of this year to gain some insights into the State of Solar in California. Over the next several days we will be reporting on what we have learned - and there are some very surprising things in here to be sure!
A word first about how we processed the CSI data. We downloaded the most recent active data set as of this writing (the August 24, 2011 data set to be precise) and parsed it into Excel. Since we were only concerned about systems in our service area, we excerpted out just the data from SCE. To narrow our focus more, we wanted to only look at applications that had significant status during the first half of this year. The CSI data has a host of date fields - we took the latest of the fields ranging from First Reservation Date to First Completed Date as our Status Date and excerpted those that fell between 1/1/2011 and 6/30/2011 - a total of 6,306 data points.
That’s a fair amount of data but it necessarily omits any data at all from the municipal utilities such as Pasadena Water & Power (where we do much of our work) or LADWP. Unfortunately, none of the munis make their program data generally available - which is particularly odd given that the local residents actually own those utilities (and thus, their data) - but that is a topic for another day.
Finally, for the purpose of these posts, all system sizes are reported in CSI Rating AC Watts (to account for differences in equipment choice and system design) as opposed to DC (or nameplate) Watts.
What can we say about those 6,306 projects? Collectively they account for 164.7 MW of new solar power at a total installed cost of just over $1 billion - with incentive amounts totaling $219 million - roughly 21% of the installed cost. Unfortunately, not all of those are built - or even ever will be. Fully 11% (698) of those projects have the status ‘Delisted’ - meaning that they have been cancelled for one reason or another. Those delisted projects account for 37.8 MW of potential solar power that presumably will never see the sun. (Do some installation companies have a significantly higher rate of “delisted” systems? We will answer that question in a subsequent post - stay tuned!)
The remaining 5,608 are split between “Installed” and “Pending” with 55.8% (3,131) installed and 44.2% (2,477) pending. Breaking that down a little more, the installed projects account for 33.8 MW worth $240.1 million with incentive amounts totaling $57.1 million. In contrast, the pending projects account for almost three times as much capacity at 93 MW worth $575.8 million with incentive amounts totaling $120.6 million. (That is, nearly three times the to-be-installed solar cpacity for roughly twice the rebate dollars.) On average, installed projects cost $7.09/Watt whereas pending projects cost $6.19/Watt - a positive trend for consumers since it shows the cost of solar power systems declining over time.
Finally, for today, let’s examine whether the data supports the notion of solar economy of scale - that is, as system size increases does the installed cost/Watt decline? To get a handle on that, we took two different cuts through our data set - “small” installed or pending systems <10 kW, and “large” systems ranging between 10 kW and 1MW.
First, here’s the graph for the “small” systems (consisting of 4,992 installed or pending systems - click on the graph to view full size). As the trend line makes clear, larger systems really do drive down costs - decreasing from over $10/Watt at the small end of the range to just above $6/Watt for systems around 10 kW.
Another interesting observation from this graph are the outliers - with some data points below $3.00/Watt (mostly from self-installed system) all the way up to nearly $18/Watt!!! (We will have way more to say about those data points - and who is responsible for them - later in this series.)
If we now look at larger systems - those between 10 kW and 1MW - our data set has 587 such systems and again, the trend line shows the decline in system costs as system size increases. (Note, because there is such a huge range in system sizes on this graph, we plotted the system size on a log scale.) Some of these outliers are also pretty curious - a 200 kW system coming in at over $14/Watt?
Of course, this data is showing what happens when an individual project gets larger and there the trend is clear. One might well ask, does the same trend apply to larger installation companies? In other words, as a company has more and more installs, does that economy of scale translate into lower costs for the end consumer? That’s a very interesting question and the answer - coming in our next post - just might surprise, or maybe even disturb you.
If there are some other cuts of this data that you would like to see, just let us know in the comments. Trust me, we are just getting started!