Notwithstanding the thunderous applause that solar photovoltaic (rooftop) power receives from enlightened opinion, it is not cost-competitive with conventional electricity, and cannot survive without massive subsidies.

An army of lobbyists, commentators and bureaucrats has told us for years that rooftop power will be cost-competitive with conventional electricity very soon, if it is not already, and technically could provide over 38 percent of U.S. electricity generation. (The figure for 2015 was three-tenths of 1 percent.)

At the same time, the same people argue that there should be no change in the subsidies that are bestowed upon it.

Such diametrically opposed arguments cannot be reconciled. With respect to the competitiveness assertion, Deutsche Bank argues: “Unsubsidized rooftop solar electricity costs between $0.08-$0.13/kWh, 30-40% below retail price (sic) of electricity in many markets globally.”

That range is somewhat higher than the Energy Information Administration (EIA) estimate of about $0.08 per kilowatt-hour (kWh), while it is lower than the range of about $0.11-$0.30/kWh estimated by Lazard.

But: Those figures exclude the costs of conventional capacity needed to back up rooftop solar capacity due to the unreliability of sunlight. The common argument that those installing rooftop solar systems can “exit the grid” is preposterous, because they must remain connected to the conventional system as insurance against periods during which sunlight conditions cannot satisfy their power demands.

In substantial part because of the system of “net metering” — rooftop solar customers are paid inflated prices for the excess power that they sell to the grid — those installing rooftop systems do not pay their proportionate shares of the costs of conventional capacity.

Everyone else must bear those backup costs. My own estimate for all renewables across all regional markets is about $0.37 per kWh. Suppose for rooftop solar systems in any given market, it is one-tenth of that estimate; that would add 30 to 50 percent to the cost of rooftop solar power estimated by Deutsche Bank.

The EIA estimate of the “levelized” (loosely, spread evenly over the life of the generation facilities) cost of gas-fired power is less than $0.06 per kWh. If we accept the Deutsche Bank estimates and add $0.04 as the cost of backup power, we get a range of $0.12 to $0.17. Accordingly, the “competitiveness” of rooftop solar power is deeply dubious even if we add to the cost of gas-fired electricity the EIA estimate of about $0.04 for transmission and distribution.

Note that the Deutsche Bank analysis points out that average electricity prices in the U.S. have increased by about 20 percent over the last decade despite declining prices for natural gas, due to an increase in transmission and distribution (T&D) costs, thus increasing the competitiveness of rooftop solar power.

But the rising T&D costs are due in substantial part to the subsidies and guaranteed market shares bestowed upon wind and non-rooftop solar power, which impose disproportionate transmission costs because of limitations on where the facilities can be sited.

That is why proponents of rooftop solar systems continue to emphasize the crucial importance of “incentives” (subsidies) for such electricity. A new analysis calculates the subsidies: For owned systems, it is $4.15 per installed watt of capacity, or 119 percent of the cost of the system (assumed to be $3.50 per watt for a 3900-watt system). For leased systems, the subsidies total $5.67 per installed watt, or 192 percent of system cost. The net metering subsidy alone is $2.20 for both owned and leased systems.

That is why reductions in the subsidies yield sharp reductions in the market for rooftop solar systems. The Nevada Public Utilities Commission about a year ago increased the fixed charges that rooftop solar customers must pay for access to the power grid, and reduced the prices paid those customers for excess power.

The major solar firms promptly exited the Nevada market, and local installers of the systems cut their operations sharply.

Why is it that such massive subsidies have failed to result in a larger market share? One basic problem is the unconcentrated energy content of sunlight. Solar energy at the top of the atmosphere is about 1,360 watts per square meter; since few of the sun’s rays are perpendicular to the earth, and because only half of the earth is illuminated at a given moment, the effective solar irradiance at the top of the atmosphere is one-fourth of that, or about 340 watts per square meter.

Moreover, because of albedo effects (the portion of solar energy reflected back to space), average solar energy at the surface is about 240 watts per square meter, of which about 20 to 30 percent is convertible to electricity with current technology. A square meter of rooftop photovoltaic panels is sufficient to power an ordinary light bulb.

New installations to a significant degree will take place in areas and regions less suitable for solar power, as the industry for obvious reasons expanded initially by exploiting the areas with the best sunlight conditions first. It is far from clear, therefore, that costs per kWh will decline even if we assume that the claims of improving scale economies and technology will prove correct.

As the old joke goes: Whom are you going to believe? The experts and industry insiders and interest groups? Or your own eyes?

The rooftop solar sector is very small despite massive subsidies and despite its purported cost competitiveness. The time has come to allow market forces to work.

This piece was originally published at the American Enterprise Institute.