The Arithmetic of Solar Royalty Trusts

Government bondsIn many ways, owning a PV system should be a dream investment. It promises dependable cash flow for an indefinitely long period of time, even a century, with little underlying market risk. It could be as good a definition of ‘risk free’ return as government bonds. Government bonds still face the challenges of national default; and PV, the challenge of a price drop in electricity. Maybe the PV is more dependable?

The basis for this is the two almost unique aspects of PV systems that are hardly ever fully exploited or even understood:

It uses no fuel, needs no on-site labor, and may have no moving parts (and thus has negligible operating costs)
The basic parts either last indefinitely (the PV module that converts sunlight to electricity) or can be replaced with nominal maintenance costs (the inverter that takes DC electricity and makes it ready for the AC grid).

An open question is how long PV modules will actually last. Modules and systems are warrantied for 30 years now, and there is discussion of 40 years. Few ask for longer, because they can’t imagine anything being used longer; but it isn’t because they believe it can’t be done. Old systems exist with modules that are almost 40 years old, and most are not dead. Most make electricity almost as well as they did originally, although this needs further verification. We can speculate that if people cared enough, PV systems could be designed to last a century.

How many investments last a century and produce a valuable commodity (electricity) at a tiny operating cost (let’s say 1 c/kWh), without inflation risk?

Why haven’t we been able to invest in PV systems? Because until recently, they have been too expensive (without complex tax credits) to make a suitable return. We could have bought such a system, but we would have been stuck making 1%-2% return on our capital. That’s why until now, you have never heard of this possibility.

But all that is changing. PV prices dropped about 40% in the last two years.

This article is the arithmetic of making money from an investment in some ways as risk free as the US government – or maybe, as risk free as Johnson and Johnson, because I hear their paper is now more expensive than the USA’s.

So let’s understand the problem. We buy a share of a PV system. Then the system turns on and produces electricity. Every year, it produces about the same amount of electricity (within 3% or so) PV modulesand costs a tiny amount of money to maintain, even including washing the modules once in awhile. Probably the price of electricity won’t tank (unless everything in the world does), and more likely it will rise and we will get paid more. The PV modules will very slowly and very slightly lose power, maybe 0.5% per year, but the price of electricity is likely to rise faster – or we can add that many more modules each year to keep output steady. And this will go on indefinitely, at least 40 years, and maybe 100.

Today, large PV systems can be built for about $3/W. (I want to thank Maja Wessels, VP at First Solar, for yesterday confirming at our 2nd annual solar symposium, that this is a sensible price for today’s large systems.) This means, if they are 10 million watts in size (10 MW), they would cost $30M. If you got a 5% return (30 year US bonds are about 5%), you would need to get $1.5M back each year. Let’s see whether you would get this money from a PV system.

Elsewhere (http://thesolarreview.org/2009/10/16/pv-fast-facts/ ), I have explained how you can figure out how much electricity a PV systems puts out per year. You will try to build your solar system where you get the best return, which means the right combination of sunlight, local price of electricity, access to the grid, system siting costs, and incentives. But in my example, I will ignore some complexity and calculate based on sunlight, system cost, O&M, annual degradation, revenue, and the 30% Federal investment tax grant.

A simple, low-maintenance non-tracking array in the Southwest can produce about 1.8 kWh per installed DC watt every year. So a 10 MW system, for example, might produce about 18 million kWh a year. Are we close? 18 million kWh at 10 c/kWh is about $1.8 million. That’s higher than the $1.5M we need to make 5%, so we are in the ballpark.

Let’s be a little more precise and practical, because we don’t want to faked out. After all, there are losses, and the losses hurt returns. It turns out, we are on the hairy edge, but that shouldn’t be a surprise. That’s why this is new.

Let’s do this all “per watt” so we can see it very clearly. So we get 1.8 kWh/W each year from our PV system. It costs us $3/W to own the system. Our annual O&M is about 1 c/kWh. We might have a transmission cost. Probably the lucky owners of the first systems will find a way to get near their load and avoid this, but let’s not assume this. Let’s assess a 1 c/kWh transmission transmission cost so we are safe.

So on a system cost per watt basis, we get the following:

Rate of Return (including capital, O&M, sunlight, transmission; but without rebates, degradation, or siting or grid connection costs or delays)

$3.5/W $3/W $2.5/W $2/W
6 c/kWh 2.1% 2.4% 2.9% 3.6%
8 c/kWh 3.1% 3.6% 4.3% 5.4%
10 c/kWh 4.1% 4.8% 5.8% 7.2%
12 c/kWh 5.1% 6% 7.2% 9%

The systems in the bottom right, low cost and high electricity price, make the 5% rate easily.

How do we estimate the impact of degradation and costs and delays of siting and grid interconnection? Current PV systems are warrantied at under 1% loss per year, but good ones are in the 0.2%-0.5% loss range. Let’s assume we replace the equivalent of 0.5% of our system output per year by buying and installing that much new PV – so the output is unchanged. That would add 0.5% of our capital cost per watt to our annual cost. How much is this per kWh? For today’s best $3/W system, this would be 1.5 c/W annual additional cost to replace 0.5% of the output. We get 1.8 kWh/yr from each watt, so this is 1.5 c/1.8 kWh, or about another penny a kWh of lost revenue. Not huge, but not insignificant. (Note that this will get smaller as system prices continue to decline.) So let’s re-do Table 1 with degradation included. It makes about a 10% difference in rate of return, which is unpleasantly significant. Stability counts.

Rate of Return with capital, O&M, sunlight, transmission and replacements for 0.5% degradation (no rebates, or siting or grid connection costs or delays)

$3.5/W $3/W $2.5/W $2/W
6 c/kWh 1.8% 2.1% 2.6% 3.3%
8 c/kWh 2.8% 3.3% 4.0% 5.1%
10 c/kWh 3.8% 4.5% 5.5% 6.9%
12 c/kWh 4.9% 5.7% 6.9% 8.7%

The 5% barrier is surpassed in the case of the highest price for electricity and for the lower price systems. But even 10 c/kWh electricity and $3/W systems; and 12 c/kWh and $3.5/W systems are close. Remember, these systems include the replacement of all lost power every year with new modules sufficient to produce that power, and at today’s prices. Those prices will drop (not accounted for), nor will they be as high with all the other preparations already in place.

The other factors are still significant. Due to the immaturity of the US market and the multiplicity of states and regulations involved, there are siting and grid connection delays. However, there are also compensating rebates and tax credits.

Perhaps the simplest way to conclude is to say, the tax credits and rebates can change a $3/W system to a $2/W system (30% tax credit is 90 c/W less); and the siting and grid connection can cost some of that advantage. But at reasonable prices for daytime electricity (around 10 c/kWh), the implied approximately $2.5/W system (after rebates but with siting costs) would meet the 5% rate of return we are seeking, with a little leeway for bumps along the way. And this completely ignores likely future increases in electricity prices and thus in PV system revenues.

Finally, how does this revenue get to the investor? If this were set up as a private partnership, like an S partnership, the income would come to owners, who would pay taxes on it. That would be great, but how many individual taxpayers can be part of complicated S corporations? If it were through a regular corporation, like those listed on the stock exchange, the corporation would have to pay taxes on the income (somewhat offset by depreciation) and then the resulting dividend would be taxed again. Individual investors would not see an adequate return and would not invest.

But protected investments like REITs, annuities, and royalty trusts pass through their income without taxation, and the owner – us, the individual investors – gets the untaxed income. If our lawmakers can make a few tweaks in the existing laws (e.g., making revenue from PV like rent, or allowing a REIT to rent a PV system to someone else who gets the actual payment for the electricity; and making the investment tax credit into a grant or rebate for this purpose), then we can get going. We can have one of the best engines of growth possible for our collective future.

Your kids will thank you if you give them one for Christmas.

Ken Zweibel

15 Comments

u34
Vasilis Fthenakis
April 21, 2010 12:12 pm

This is an incredible message. Since we make conventional power plants with hundreds of moving parts last 40+ years, there is a certainty that we can make PV power plants lasting even longer and have them providing almost free energy after their pay-off. Add the environmental benefits, and we have a cost justification for going ahead with very large deployment right now.

u28
joe1347
April 22, 2010 6:03 pm

Question, assuming that PV module costs continue to drop fairly rapidly (as PV technology improves), how would that effect the value of a PV REIT purchased today? Would the value of the PV REIT drop in value because new PV installations are much more cost effectitive? Granted, the rate of return might stay the same, but wouldn’t the trading value of the RV REIT drop. I’m guessing that one of the appeals of REIT’s to investors (i.e., one reason why they might buy or want them) is the hope that they might eventually go up in value. With a PV REIT – are you pretty much guaranteed that they’ll ALWAYS lose value over time as more efficient (lower cost) solar systems become available?
As for the 40 year PV module lifetime, has there been any thought given to being able to rebuild or rework the silicon-based PV modules after they’ve ‘degraded’. By degraded, I’m guessing that the silicon solar cell substrate itself is only marginally degraded (if at all), instead the metallic interconnects may be corroded and the module seals may be compromised. Could you remove the silicon solar cells from a compromised PV module and then put them in a new module and if necessary, just redo the emitter metallization. Not a ‘free’ solution, but likely much cheaper than replacing the entire module (with a new module)?

u6
gwsolar
April 23, 2010 9:35 am

Isn’t it more likely that you will have to pay more to buy into the newer systems, so that the return stays the same? I believe the price of the asset will vary with the relative return to similar assets. Meanwhile, if the underlying price of electricity goes up, so will the value of the asset.
The first thing that will happen is that the incentives will be phased out, so the price relative to rate of return first 10-15 years will probably be the same. In other words, if $2/W systems after rebate of the investment tax credit are attractive, the rebate will drop steadily as the cost of the system drops towards $2/W.
But this is just a first impression. Perhaps not germane to the attractiveness of the concept.
Recycling cells and materials is possible, and some have suggested replacing old modules with new, cheaper more efficient ones. That will simply be a strategy, if it makes sense. I am not convinced that tearing out working modules for better ones (and adding capital cost to do so), will make sense unless the land becomes more valuable than the system (i.e., the region needs the land for another purpose).
This is more detail than I want to express, given the uncertainties.

u39
Tom
May 9, 2010 3:35 am

A new PV REIT will have have higher value than an old one… if a PV REIT with a 40 year lifetime has already been around 10 years, it will only have 30 years of life left, and hence have to have lower value. The income from the REIT will have to be high enough to make up for this loss of value.
The direct comparison with a bond used in the article is misleading… bonds pay the inital investment back at the end of their useful life, while PV systems are worth very little when they stop producing electricity. Hence, a PV system would need to create a higher income stream than a bond to be an equally attractive investment.

u39
Tom
May 9, 2010 3:29 am

Ken,
I just read your April 21 article on Renewable Energy World, “The
Arithmetic of Solar Royalty Trusts.” The intent of the aritcle is
well received… given today’s low panel prices and low interest
rates, PV is starting to look good in many circumstances as a
financial investment.
However, I’m afraid your calculations are misleading, and they
make PV look better than it is. It’s not fair to compare the income
from a PV system with the income from a bond directly, because a bond
returns the initial investment at maturity, while a PV system is
worthless at its end of life. A 30 year, 5% bond will return 5% x 30
+ 100% = 2.5x the original investment over 30 years, while a PV system
that lasts 30 years and repays 5% each year will only return 5% x 30 =
1.5x the original investment.
The proper way to compare PV with financial investments is to use
Internal Rate of Return (IRR).
http://www.investopedia.com/terms/i/irr.asp. There is an IRR function
in most spreadsheets.

u6
gwsolar
May 9, 2010 3:31 pm

Tom,
In the calculation, I am assuming that the degradation rate is a steady 0.5% and there are no catastrophic losses. In other words, the system lasts indefinitely. In addition, I am adding back 0.5% of the system each year, to keep output the same. This makes calculations simpler, but is not essential to the concept.
The major point is the one you object to – PV systems don’t simply go to zero. They maintain their value, and in a best-of-all-possible worlds manner, actually don’t even have to be recycled – they just keep going.
The length of life of PV systems is not established. I am suggesting that this long life, now 30-40 years, can be extended if we design for it. There is certainly no counter example saying it is impossible or even unlikely.
Yes, this concept is ‘mold breaking’ (i.e., it causes existing experts to miss it if they do not read closely) and thus all the more worthwhile to seriously consider its implications.

u39
Tom
May 9, 2010 6:14 pm

Ken,
You’re right… I was skimming too much. The system you’ve designed by paying to replace degradation with additional panels is closer to a 30 yr bond than I thought… it’s actually a perpetuity, and so would be comparable to the interest rate on a very long term bond with similar liquidity and credit quality.
Adjustments for liquidity and credit risk are clearly beyond the scope of your article, so I was wrong to object.

u39
Tom
May 9, 2010 6:32 pm

By the way, using IRR on a PV system without adding panels as the system degrades would make the system look better as an investment, not worse.
This is because a degrading PV system can be fairly compared to a bond with a shorter duration, and short term interest rates tend to be lower than longer term rates, and increasing the number of panels as you are doing will increase rates.

u6
gwsolar
May 9, 2010 9:30 pm

I am trying to explore the ‘space’ of perpetual, dependable investments because I am skeptical of the ideals of current economic theory of discount rates. I believe they are part of the mesmerization of current hopes rather than fact. They degrade the value of dependable, low risk investments in favor of speculation – recently exposed as the basis of wall street wealth.

u39
Tom
May 10, 2010 12:03 am

Regarding discount rates, they do lead to some strange behavior, but I don’t think we can throw them out. They’re more than economic theory… they are embedded in human psychology.
Unless you can honestly say that you would be indifferent between receiving $100 today and $100 ten years from now, it’s too early to throw out the concept of discount rates. Discount rates apply even when we’re not talking about money… would you rather remove a ton of CO2 from the atmosphere today, or in 2050?
I have a problem with discount rates, too: I believe that our psychology leads us to use discount rates that are much too high to be good for our long term well-being as a species… but this is not a problem with the economic theory, it’s a problem with human psychology.

u6
gwsolar
May 10, 2010 10:42 am

I think mainstream thinking is stuck with some pretty rigid ideas of what is of value. What if you asked me, which would you rather have, $10 now when your tax rate is high, or $10 after you retire, and it isn’t? I think we have a lot of assumptions built into our models that are like the ‘efficient market’ – they simplify, and during normal periods, do not set off alarms. But they have no factual content. They are simplifying assumptions that fail outside their domain, and that domain is much less broad than people think or act on.
In any case, I agree with the bottom line – we use discount rates for societal choices that are too high.

u39
Tom
May 10, 2010 2:21 pm

I think you’re right about simplifying assumptions in models leading to the wrong conclusions. My biggest problem with the discount rate model is that it uses a fixed discount rate. Behavioral economics studies have repeatedly demonstrated that our psychological discount rates are much quite high in the near term, while they drop to near zero in the long term.
For example, we have only a slight preference between receiving a dollar 10 years or 11 years from now, but we care a lot if we get it today, or a year from now.
Since most economic theory assumes discount rates are constant, it often underestimates short term discount rates, and always overestimates long term discount rates. Because of this, it will always put too low a value on the very long term (30 years+) and sometimes undervalues the very short term.
My intuition is that a real human discount rate is a lot closer to something like R/n or R/n^2 for year n. R will be between 10% and 50%, depending on the individual (the poor and people who feel less safe tend to have higher discount rates.) The economically sophisticated probably have discount rates that are closer to the economic models (just R), but I believe that that is because of the influence of the models on our thinking.
I think your perpetual PV systems would look better in comparison to more typical financial instruments with this sort of declining discount rate, even if the first year discount rate is higher.