How the Lights Stay On

Chapter 9: The Bare Roof

I’ll be honest. The first time I really understood how solar panels worked, I started looking at every roof I drove past with a small, irritated voice in my head.

“Why isn’t that one covered?”

It was hot. It was sunny. The sun was free. Solar panels keep getting cheaper. Anyone with a roof should be able to put a few panels up and shave a chunk off their electric bill, right?

For years, I assumed that any roof that did not have solar panels on it must belong to someone too lazy, too disorganized, or too uninterested to do the obvious thing.

I was wrong. The story is more interesting than that.

This chapter is about the bare roof. About all the reasons a roof you would expect to have solar might not. About why “just put up some panels” is one of those phrases that sounds simple until you actually try to do it.

Most of all, this chapter is about a deeper truth that comes up over and over in this book: rooftop solar is not really a homeowner question. It is a grid question.

Let’s start with the dream. Then we’ll work through what makes it harder.

The sun delivers, on average, about a kilowatt of power per square meter to the ground in a sunny place. A typical American house has 1,500 to 2,500 square feet of roof. Even if you only cover half of that with panels, even at modern panels’ modest 22 percent efficiency, the math works out. A well-positioned American home could generate, over the course of a year, more electricity than the family inside it uses.

The hardware to do this has fallen in price by more than 90 percent in the last fifteen years. A 6-kilowatt system that cost $50,000 in 2010 might cost $15,000 today, before tax credits. That is a real, dramatic price drop. There has been no comparable price drop in any other major piece of household equipment, ever.

So the dream is: a sunny roof, modern cheap panels, free sunlight. The homeowner makes their own electricity, sells any extra back to the grid, and watches the bill shrink.

That dream is not impossible. Millions of American homes have it. In some neighborhoods of California, Arizona, Hawaii, Texas, and Florida, the majority of single-family houses now have solar. The dream works.

But the bare roof next door to a solar-covered one is not necessarily owned by someone foolish or lazy. It is more likely the product of a small stack of real constraints, any one of which can be the dealbreaker.

Let’s walk through them.

Not all roofs are good roofs for solar.

Orientation matters a lot. A south-facing roof in the United States gets the most sunlight. East and west are okay. A north-facing roof is, for solar purposes, almost useless. If your house faces the wrong direction, the big obvious-looking roof from the street is actually the back of the house, and the back of the house faces north.

Shade matters even more. Solar panels are surprisingly intolerant of shade. A single tree branch shading a single panel can drop the output of a whole string of panels significantly. Newer panels with built-in optimizers help, but they cost more. A neighbor’s tall oak tree, planted thirty years ago by someone who did not anticipate solar panels, can make the whole project not worth doing.

Roof age matters too. Solar panels are heavy. They last 25 to 30 years. If your shingles have only 10 years of life left when the panels are installed, you are about to have a painful problem. You will either have to remove the panels, reroof, and reinstall them — adding thousands of dollars to the project — or live with leaks. Most installers will tell you that if your roof has less than 15 years left, you should reroof first and install solar second. Adding $10,000 to a $15,000 project changes the math considerably.

Roof material matters in a smaller way. Asphalt shingle roofs are easy. Tile roofs are harder. Metal roofs are easy. Wood shake roofs are a fire risk and may not be permitted with panels at all.

So just from the roof itself, before any of the other stuff, a meaningful fraction of American homes are simply not great candidates for solar. The neighbor with the bare roof might have a north-facing house, or a roof from 2003, or an oak tree.

Here is a constraint nobody thinks about until they have lived through it.

The person who owns the roof is often not the person who pays the electric bill.

About 36 percent of American households rent. Renters do not own their roofs. They cannot install solar panels even if they want to. Their landlord could install panels, but the landlord does not pay the electric bill, so the savings would go to the renter, not to the landlord. The landlord pays the construction cost and gets nothing back. So the panels do not get installed.

This is called a split incentive in economics. The person who could pay for the upgrade does not benefit from it. The person who benefits cannot pay for it. The economically optimal outcome — install the panels, share the savings — requires a complicated arrangement that almost never gets made in practice.

The same split-incentive problem affects most commercial buildings (the building owner is not the tenant), most condominiums (the homeowners’ association has to agree), and a lot of multi-family housing.

In some states and cities, programs exist to bridge this gap. Community solar lets renters and people with bad roofs buy a share of a solar farm somewhere else, and get the credit on their electric bill at home. It works. But it is much less common than rooftop solar and tends to be a smaller economic benefit.

The bare roof down the street might belong to a landlord who would be financially crazy to install panels they cannot profit from.

Now we get to the part that determines, more than anything else, whether a sunny roof in a sunny state has solar on it.

The rules that decide what happens to the electricity your panels generate.

A solar panel system on a house is connected to the grid. During the day, when the sun is up and the panels are producing, the system may generate more electricity than the house is using at that moment. The extra has to go somewhere. It flows back out through the meter, onto the wires in the street, and into the grid.

The big question: does the utility pay you for that extra electricity? And if so, how much?

For most of the last twenty years, the answer in the United States was: yes, the utility pays you, at roughly the same rate it would charge you for buying it. This is called net metering. Your meter literally spins backward. If you generated more electricity than you used over a billing period, the utility owed you money.

Net metering was the policy that made rooftop solar economic for ordinary households. It is the reason California, Arizona, Massachusetts, and New Jersey became solar leaders.

But net metering has a problem. From the utility’s point of view, it is paying retail prices for electricity it could buy on the wholesale market for a fraction of that. Worse, it is paying retail prices for electricity that arrives mostly at noon — when, as we learned in Chapter 6, there is already too much solar on the grid and prices on the wholesale market are sometimes negative. The utility is buying noon-sun at dinner-time prices, then trying to sell it to other customers at dinner time too. The math does not work for the utility.

So utilities have, for over a decade, been pushing to reduce or eliminate net metering. The fight has played out state by state, regulator by regulator.

In California, the fight came to a head in 2023. The California Public Utilities Commission replaced the previous net metering policy with a new one called NEM 3.0 (Net Energy Metering, version 3). NEM 3.0 took effect on April 14, 2023.

Under NEM 3.0, a California homeowner who exports a kilowatt-hour of solar to the grid is no longer paid retail rate. They are paid a much smaller amount based on something called the Avoided Cost Calculator — essentially, the wholesale value of the electricity to the grid at that exact moment. For solar exported at noon, when the grid is flooded, this rate can be 70 to 80 percent lower than the retail rate.

A California homeowner who used to get about thirty-six cents per kilowatt-hour for exported solar now gets about eight cents.

The effect was immediate. Rooftop solar sales in California fell by between 66 and 83 percent in 2023 compared to 2022. By the end of 2023, the state had lost about 17,000 solar-industry jobs — 22 percent of California’s solar workforce. Thirty-two solar companies in California went out of business in 2023 and 2024, including ADT Solar, one of the country’s largest.

Three years later, in March 2026, after multiple lawsuits, the California Court of Appeals upheld NEM 3.0. The policy is, for now, the law.

Different states have different rules. Many states still have something close to full net metering. Some have moved toward versions of net billing similar to NEM 3.0. Arizona moved away from full net metering in 2017. New York and Massachusetts still have strong programs. Florida tried to weaken its program; the governor vetoed the change. The map keeps changing.

The honest summary is this: two identical houses, on identical streets, with identical sunlight, can have wildly different solar economics depending on which side of a state border they’re on. And those state borders are themselves political, not natural. A roof in San Diego does not have solar for very different reasons than a roof in Phoenix.

There is one more constraint that almost nobody thinks about, and it deserves its own moment because of how surprising it is.

Sometimes the local distribution transformer — the gray canister on the pole or the green box in the yard, from Chapter 5 — cannot accept any more solar.

Here is why. A neighborhood transformer was designed to send electricity one way: down from the wires on the pole into the houses. When solar panels are exporting electricity back up through the transformer at noon, the transformer is being used in a way it was not originally designed for. There is a limit to how much backflow it can safely handle.

When that limit is reached, the utility will not approve more solar installations on that transformer’s circuit. The houses on that circuit are simply told no. The utility has to upgrade the transformer first, which costs money and takes time, and which usually does not happen until enough houses on the block are demanding it.

This is happening more and more in older neighborhoods with high solar adoption. The bare roof next door to four houses with solar might be bare not because the homeowner does not want panels. It might be bare because the utility said no, and the homeowner does not know how to fight that fight.

Add to all of this the constraint we already met in Chapter 5: the service panel and wiring inside the house.

A solar installation requires that the house’s electrical panel can accept the connection. Old houses with 60-amp or 100-amp service may need a panel upgrade before solar can even be installed. A panel upgrade can cost three to seven thousand dollars and take months to schedule because of the shortage of electricians in many parts of the country.

Add this to the cost of the solar system itself, and the payback math can shift from “ten years” to “twenty years.” Most people will not invest in a twenty-year payback unless they plan to stay in the house for at least that long.

The bare roof might be bare because the panel inside the basement is original to a 1962 build.

So we have a stack.

The roof itself (orientation, age, shade). The owner-occupant mismatch (renters, condos, commercial). The state policy (net metering rules). The local transformer (saturated or not). The house’s own wiring and panel.

Any one of these can be a dealbreaker. Most bare roofs are bare because at least one of them applies. A surprisingly large number of bare roofs have all of them applying at once.

This is what I mean when I say rooftop solar is a grid question, not just a homeowner question.

The choice to put solar on your roof is partly yours, yes. But it is also a choice by your state regulator about how exported solar is compensated. It is a choice by your utility about how much they have invested in the local transformer. It is a choice by your landlord, or your spouse, or your homeowners’ association. It is constrained by the age of your roof and the angle of the sun and the trees your neighbors planted forty years ago.

When you see a bare roof, you are not necessarily looking at a failure of individual will. You are usually looking at a system that has not yet been built — or, in California’s case, a system that was rolled back.

Rooftop solar is genuinely good. The sun is free. The panels work. Millions of American homes are now generating most of their own electricity. In sunny states with good policy, the payback is reasonable and the climate benefit is real.

But rooftop solar is not the whole answer to a clean grid. It is one piece of a much larger system that includes utility-scale solar farms in the desert, utility-scale wind farms on the plains, transmission lines linking them, storage to buffer them, and the grid management we have been learning about throughout this book.

In many places, utility-scale solar is now cheaper per kilowatt-hour than rooftop solar — meaning it is more efficient to build big solar farms in the desert and ship the electricity to your house than to install panels on your roof. The rooftop is more visible, more empowering, and more local. The desert farm is cheaper and more efficient. Both are true.

The bare roof, in other words, is not always a problem to be fixed. Sometimes it is the rational outcome of a system that has decided, correctly, that the electricity is better generated somewhere else.

The next time you drive through a sunny neighborhood, you can look at the bare roofs differently.

Some of them are bare because the homeowner has not gotten around to it. Some are bare because the homeowner is a landlord whose tenants pay the electric bill. Some are bare because the state changed the rules. Some are bare because the transformer on the pole next door is already full. Some are bare because the wiring inside is older than the moon landing.

And some of them are bare because somewhere out in the desert, a solar farm has already been built — to make electricity more cheaply than the bare roof ever could — and a transmission line is bringing it to that very house.

The roof is just one place where the choice gets made. The choice itself happens at a much bigger scale.

All 18 chapters of How the Lights Stay On · Library · All courses