Payback and incentives after the federal credit
Is solar a better investment than the stock market? An 8 kW system's return against 74 rolling 25-year windows of the S&P 500
At the Department of Energy's $2.95 per watt benchmark, an 8 kW rooftop system whose output is worth 18 cents a kilowatt-hour returns about 5.7 percent a year over 25 years, with no tax credit counted and the DOE running-cost allowance paid out of the savings. Across the 74 rolling 25-year windows from 1928 to 2025, the S&P 500 with dividends returned a median of 10.3 percent a year, so on that roof solar needs each kilowatt-hour to be worth about 25 cents to match the median.
The comparison is closer than the median suggests. The worst of the 74 stock windows, 1929 to 1953, still earned 5.2 percent, and the median window for 10-year Treasury bonds earned 5.0 percent, so a roof at 18 cents lands half a point above the worst stock window and 4.6 points below the median one. Move the value to 12 cents and the same roof returns 1.1 percent; move it to 30 cents and it returns 13.0 percent. The value of each kilowatt-hour is the input that decides, which is why the tables below use it as the axis instead of a state or a rate plan.
Leaving $23,600 in the index and withdrawing every year what the panels would have saved gives a second test, and it is harsher on stocks than the averages are. At 18 cents the index paid all 25 years of withdrawals in 64 of the 74 windows and ran out in the ten that began in 1928 to 1931, 1937, 1966 and 1998 to 2001. Every stock figure is history from a stated series, and every solar figure is arithmetic on inputs printed beside it. None of it is a forecast of stock returns or of electricity prices, and none of it is investment advice.
The benchmark system and its cash flows
The Department of Energy's cost benchmark supplies the price. Table 2 of the 2025Q1 PV-only benchmarks gives the residential 8 kW DC system a modeled market price of $2.95 per watt DC and a minimum sustainable price of $2.78, with operations and maintenance of $34 per kW DC per year, all in 2024 dollars. DOE states that its benchmark costs "do not include subsidies payable to the system owner, which reduce the owner's effective cost." Eight kilowatts at $2.95 is $23,600. The DOE page still lists the 2025Q1 set as its current benchmark on November 4, 2026, and this site treats the figure as a comparison line, not a quote.
Each year's cash flow is the system's output times the value of a kilowatt-hour, minus the running-cost allowance. Year-1 output is 8 kW times a production factor, and the tables use 1,300, 1,500 and 1,700 kWh per kW per year, which is 10,400, 12,000 and 13,600 kWh. They are planning factors for a roof without a shade problem; the production-estimate guide shows how to replace them with a roof-specific run. Output then falls 0.5 percent a year, the rate DOE's Federal Energy Management Program gives when it says PV module efficiency "unavoidably degrades at about 0.5% per year." Year 25 therefore delivers 88.7 percent of year 1. The horizon is 25 years, inside the "20- to 30-year lifespans" the same program gives for PV systems.
The $34 per kW running-cost line is $272 a year for 8 kW, held flat in nominal dollars here. DOE says the residential line "includes the same cost elements that apply to" the utility-scale system, which it lists as module cleaning, periodic inspection, repairs, replacement of components, property tax, insurance and management, without the land lease. It adds that the model includes a cost for the homeowner's own time on cleaning and inspection, "although most homeowners do not directly acknowledge this cost." On the other two items it says: "The cost of increased property tax and insurance coverage are also included in the model, although these costs are often not assessed to the property owner." An owner who pays none of that has a better return than the net columns show. The return table therefore carries one column before the allowance. Because the allowance covers replacement of components on average, no separate inverter line sits in the model; the inverter replacement guide prices that second purchase on its own.
The input that moves everything is the value of a kilowatt-hour, and it is a blend. Energy the house uses as the panels make it replaces power bought at the retail price, and energy sent to the grid earns whatever the export rule pays. The net metering explainer sorts the export rules and the break-even guide turns a value into a payback. For scale, the Energy Information Administration's Electric Power Monthly, Table 5.6.A (released October 23, 2026), puts the U.S. average residential price for August 2026 at 18.16 cents per kWh, against 17.57 a year earlier, with Nevada the lowest state in the table at 12.72 cents and Hawaii the highest at 47.18. Those are prices for power a household buys. A kilowatt-hour exported under a net billing tariff earns less, and the causes of rate movement are EnergyBS's subject, so the EnergyBS electricity rate guide holds that side. The tables here run from 8 to 35 cents so any household can find its row.
| Input | Value used | Where it comes from |
|---|---|---|
| System | 8 kW DC, residential PV only | DOE Solar Photovoltaic System Cost Benchmarks, 2025Q1, Table 2 |
| Price paid | $23,600 ($2.95 per watt DC modeled market price, 2024 dollars) | DOE Table 2; used as a comparison line, not a quote |
| Federal tax credit | None | IRS: not available for property placed in service after December 31, 2025 |
| Running-cost allowance | $272 a year ($34 per kW DC per year), flat in nominal dollars | DOE Table 2 O&M; the flat nominal treatment is this guide's |
| Output in year 1 | 10,400, 12,000 or 13,600 kWh (1,300, 1,500 or 1,700 kWh per kW) | Planning factors for an unshaded roof; replace with a roof-specific run |
| Output decline | 0.5 percent a year, compounding | DOE Federal Energy Management Program |
| Horizon | 25 years, no resale or residual value | Inside the 20 to 30 year lifespan DOE FEMP gives |
| Value of a kWh | 8 to 35 cents, flat unless a column says otherwise | The axis of the tables; EIA anchors below |
Annual return by value of a kilowatt-hour
Return in this guide means the internal rate of return: the single annual rate that makes the 25 net cash flows, discounted, equal the $23,600 paid at the start. It is nominal, before tax, and it assumes each year's saving arrives at year end. Simple payback, covered in the payback section, divides the price by one year's saving; return keeps all 25 years, which is what lets it sit beside an index. The table computes it for nine values and three output levels, with the 1,500 column repeated before the running-cost allowance.
| Value of a kWh | 1,300 kWh per kW, net of DOE allowance | 1,500 kWh per kW, net | 1,700 kWh per kW, net | 1,500 kWh per kW, before the allowance | Simple payback at 1,500 (years) |
|---|---|---|---|---|---|
| 8 cents | -4.4% | -2.9% | -1.7% | -0.3% | 34.3 |
| 10 cents | -2.2% | -0.7% | 0.5% | 1.5% | 25.4 |
| 12 cents | -0.3% | 1.1% | 2.5% | 3.1% | 20.2 |
| 15 cents | 2.0% | 3.5% | 5.0% | 5.3% | 15.4 |
| 18 cents | 4.0% | 5.7% | 7.3% | 7.2% | 12.5 |
| 20 cents | 5.2% | 7.0% | 8.7% | 8.5% | 11.1 |
| 25 cents | 8.1% | 10.1% | 12.0% | 11.5% | 8.7 |
| 30 cents | 10.7% | 13.0% | 15.2% | 14.3% | 7.1 |
| 35 cents | 13.2% | 15.7% | 18.2% | 17.0% | 6.0 |
Annual return at selected values per kWh and three output levels.
| Value of a kWh | 1,300 kWh per kW | 1,500 kWh per kW | 1,700 kWh per kW |
|---|---|---|---|
| 10 cents | -2.2% | -0.7% | 0.5% |
| 15 cents | 2.0% | 3.5% | 5.0% |
| 20 cents | 5.2% | 7.0% | 8.7% |
| 25 cents | 8.1% | 10.1% | 12.0% |
| 30 cents | 10.7% | 13.0% | 15.2% |
| 35 cents | 13.2% | 15.7% | 18.2% |
Each extra 5 cents adds roughly 3 points at 1,500 kWh per kW: 3.5 percent at 15 cents, 7.0 at 20, 10.1 at 25 and 13.0 at 30. Below about 10.8 cents the 25 years of net output do not repay the price at all; at 10 cents they sum to $21,467 against $23,600, a return of minus 0.7 percent. Output matters about as much as value. At 18 cents the return runs from 4.0 percent on a 1,300 kWh roof to 7.3 percent on a 1,700 kWh roof, so a 1,300 kWh roof at 20 cents (5.2 percent) earns about what a 1,700 kWh roof earns at 15 cents (5.0 percent). The running-cost allowance takes between 1.3 and 2.0 points off the return from 12 to 30 cents, which is the gap between the last two return columns.
Two sibling tables on this site use the same ingredients, and the numbers agree. With no allowance, 12,000 kWh at 20 cents is $2,400 in year 1 and the return is 8.5 percent, the figure the cash-versus-loan guide prints for the $2.95 row of its own return table. With the DOE allowance it is 7.0 percent. Over 25 years a 1,500 kWh roof makes 282,671 kWh, so the $23,600 price is 8.3 cents per kWh and the $30,400 of price plus 25 years of allowance is 10.8 cents, the pair in the break-even guide's lifetime-cost table.
How much the price per watt matters
The return falls fast as the price rises above the benchmark, because the price is the one number paid up front. The table holds the roof at 1,500 kWh per kW and moves the price per watt, at two values per kWh. The $2.78 row is DOE's minimum sustainable price, a competitive-market floor and not a price a homeowner is promised; the $3.50, $4.00 and $4.50 rows are illustrations of quotes above the line.
| Price per watt DC | System price (8 kW) | Return at 18 cents | Return at 25 cents |
|---|---|---|---|
| $2.78 (DOE minimum sustainable price) | $22,240 | 6.3% | 10.9% |
| $2.95 (DOE modeled market price) | $23,600 | 5.7% | 10.1% |
| $3.50 | $28,000 | 4.0% | 7.9% |
| $4.00 | $32,000 | 2.7% | 6.4% |
| $4.50 | $36,000 | 1.7% | 5.1% |
Going from $2.95 to $3.50 per watt adds $4,400 and costs 1.7 points at 18 cents and 2.2 points at 25 cents. At $4.50, the 18-cent return is 1.7 percent, below the 3-month bill's median 25-year window of 3.7 percent. A quote that high needs an explanation in the scope of work, and the quote red-flag guide and the quote-versus-marketplace reconciliation show how to ask for it. The horizon also counts. At 18 cents on the 1,500 kWh roof the return is 4.4 percent over 20 years, 5.7 over 25 and 6.4 over 30; at 12 cents it is minus 0.7, 1.1 and 2.2 percent, and at 25 cents 9.2, 10.1 and 10.5 percent. A longer life helps least where the value per kWh is already high, because at a higher return the later years count for less and output keeps shrinking.
What the S&P 500 earned over 25 years
The stock side comes from Aswath Damodaran's Historical Returns on Stocks, Bonds and Bills at NYU Stern, a table of annual returns from 1928 through 2025 for the S&P 500 including dividends, the 10-year Treasury bond and the 3-month Treasury bill. The table's header still reads 1928-2024, but it carries a 2025 row and says it was updated January 5, 2026, and Damodaran notes that the bill column is now the average bill rate during the year. All three series are nominal, before tax and before any fund fee. One internal inconsistency affects the bond column: the 2025 bond return prints as 7.80 percent, while the table's cumulative-value column implies 8.29 percent. This guide uses the printed 7.80 percent; the other figure would move the 98-year bond return from 4.534 to 4.539 percent and the 2001-2025 window from 3.34 to 3.35 percent.
Over the 98 years the S&P 500 compounded at 10.02 percent a year, against 4.53 percent for the 10-year bond and 3.37 percent for the bill. Those averages hide the path. The index lost money in 26 of the 98 calendar years and fell more than 20 percent in six of them (1930, 1931, 1937, 1974, 2002 and 2008); 1931 was minus 43.84 percent, and from the end of 1928 through 1932 the cumulative return was minus 64.8 percent. The decade table shows the same unevenness at a longer scale: 19.5 percent a year in the 1950s, minus 1.0 percent a year from 2000 through 2009, a decade in which the 10-year bond returned 6.3 percent a year.
| Period | S&P 500 | 10-year Treasury bond | 3-month Treasury bill |
|---|---|---|---|
| 1928-1939 | 1.5% | 3.7% | 1.3% |
| 1940-1949 | 8.5% | 2.5% | 0.5% |
| 1950-1959 | 19.5% | 0.8% | 2.0% |
| 1960-1969 | 7.7% | 2.4% | 4.0% |
| 1970-1979 | 5.9% | 5.4% | 6.3% |
| 1980-1989 | 17.3% | 12.0% | 9.0% |
| 1990-1999 | 18.0% | 7.4% | 5.0% |
| 2000-2009 | -1.0% | 6.3% | 2.7% |
| 2010-2019 | 13.4% | 4.1% | 0.6% |
| 2020-2025 | 14.9% | -0.6% | 2.8% |
Rolling 25-year windows for stocks, bonds and bills
A 25-year window, the horizon of the solar system, smooths the path without removing it. The 74 windows run from 1928-1952 to 2001-2025, and the table gives their distribution for the three series. Their S&P 500 returns range from 5.22 percent for 1929 to 1953 to 17.11 percent for 1975 to 1999, with a median of 10.32 percent. One window fell below 6 percent, two below 7, seven below 8, 13 below 9 and 32 of the 74 below 10 percent. The bond's median window earned 4.98 percent and its worst, 1945 to 1969, only 1.86 percent. The bill's median was 3.72 percent. The percentile rows are interpolated between ranked windows.
Two cautions apply to every number in the table. The windows share most of their years, since 74 windows come from 98 annual returns, so they are not 74 independent trials; they are one history read from different starting points. And the history comes from a market whose composition and starting valuations changed over the period, so a past window range does not set a floor or a ceiling for the next 25 years.
| Series | Worst window | 10th percentile | Median | 90th percentile | Best window | All 98 years, annualized |
|---|---|---|---|---|---|---|
| S&P 500 (includes dividends) | 5.22% (1929-1953) | 8.36% | 10.32% | 13.68% | 17.11% (1975-1999) | 10.02% |
| 10-year Treasury bond | 1.86% (1945-1969) | 2.30% | 4.98% | 8.84% | 9.62% (1981-2005) | 4.53% |
| 3-month Treasury bill | 0.76% (1931-1955) | 1.05% | 3.72% | 7.07% | 7.29% (1967-1991) | 3.37% |
Set against these rows, the 5.7 percent of the 18-cent roof at 1,500 kWh per kW is above the worst S&P 500 window and above the bond return in 42 of the 74 windows, and below the index return in 73 of the 74. The more useful framing is the value of a kilowatt-hour that would be needed to match each benchmark, which the next section computes.
The kilowatt-hour value that matches each benchmark
Turning the return table around gives a hurdle: for a given output, the first-year value of a kilowatt-hour at which the benchmark system earns exactly the return a historical benchmark earned. A household that knows its output and its value can read the row it clears. All entries are net of the DOE allowance with the value held flat for 25 years.
| Benchmark | Its annual return | 1,300 kWh per kW | 1,500 kWh per kW | 1,700 kWh per kW |
|---|---|---|---|---|
| 3-month T-bill, median window | 3.72% | 17.6 | 15.2 | 13.4 |
| 10-year T-bond, median window | 4.98% | 19.6 | 17.0 | 15.0 |
| S&P 500, worst window | 5.22% | 20.0 | 17.3 | 15.3 |
| S&P 500, 10th percentile | 8.36% | 25.5 | 22.1 | 19.5 |
| S&P 500, median window | 10.32% | 29.3 | 25.4 | 22.4 |
| S&P 500, 90th percentile | 13.68% | 36.1 | 31.3 | 27.6 |
On the 1,500 kWh roof the bill median needs 15.2 cents, the bond median 17.0 and the worst stock window 17.3. The median stock window needs 25.4 cents and the 90th percentile 31.3. If every kilowatt-hour earned the August 2026 U.S. average of 18.16 cents, that roof would clear the first three rows and fall 7.2 cents short of the median stock row. On a 1,300 kWh roof the same average falls short of the bond median (19.6 cents needed), and on a 1,700 kWh roof it clears the bond median (15.0), the worst stock window (15.3) and misses the median stock window (22.4).
A rising value lowers the hurdle, because the later cash flows are larger. If the first-year value grows 2 percent a year, the 1,500 kWh roof needs 21.9 cents to match the median stock window, and at 3 percent a year it needs 20.2, compared with 25.4 for a flat value. To match the worst stock window the same roof needs 17.3 cents flat, 14.3 at 2 percent and 12.9 at 3 percent, and to match the bond median 17.0, 14.0 and 12.6.
Paying the same bills from an index fund
An internal rate of return is usually read as assuming that every dollar the system saves is reinvested at that same rate, and a stock return is not a smooth rate. A second test removes the assumption. Put the $23,600 in the index on January 1 of the first year of a window, then at the end of each of the 25 years withdraw exactly what the panels would have saved that year, net of the running-cost allowance. If the balance never goes below zero, the index paid for what the panels pay for. The test uses each window's actual annual returns in order, with no averaging, so a bad early year hurts as it would have in life. It is pre-tax and ignores fund fees.
| Value of a kWh | Solar return | Windows where the index return beat it | Windows where the S&P 500 paid all 25 withdrawals | Same, 10-year Treasury bond | Same, 3-month Treasury bill | S&P 500 median balance left after year 25 (successful windows) |
|---|---|---|---|---|---|---|
| 10 cents | -0.7% | 74 of 74 | 73 of 74 | 74 of 74 | 74 of 74 | $183,989 |
| 12 cents | 1.1% | 74 of 74 | 72 of 74 | 74 of 74 | 61 of 74 | $167,200 |
| 15 cents | 3.5% | 74 of 74 | 69 of 74 | 42 of 74 | 38 of 74 | $142,034 |
| 18 cents | 5.7% | 73 of 74 | 64 of 74 | 31 of 74 | 20 of 74 | $114,445 |
| 20 cents | 7.0% | 72 of 74 | 58 of 74 | 21 of 74 | 10 of 74 | $98,356 |
| 25 cents | 10.1% | 42 of 74 | 38 of 74 | 6 of 74 | 0 of 74 | $84,718 |
Two counts in the table answer different questions. The fourth column counts windows in which the index's 25-year annualized return was higher than the solar return, which is 73 of 74 at 18 cents. The fifth counts windows in which the withdrawals could actually be paid, which is 64 of 74. The gap of nine windows is sequence risk: in those windows the index averaged more than 5.7 percent over 25 years, but it fell early, and a portfolio that is being drawn down cannot wait for the recovery. The one window where the index's average was below the solar return is 1929 to 1953, which also failed. The ten failures at 18 cents began in 1928, 1929, 1930, 1931, 1937, 1966, 1998, 1999, 2000 and 2001. The 1966 and 1998 windows paid 24 years of withdrawals and fell short in the last.
The failures spread as the value rises. At 25 cents the S&P 500 paid all 25 withdrawals in 38 of the 74 windows and failed in 36, with starts in 1928 to 1932 and 1934 to 1940, 1956, 1957 and 1959 to 1974, and 1996 to 2001. The 10-year bond funded 31 windows at 18 cents and only 6 at 25 cents, and the bill funded 20 and 0. Where the index does pay, it pays with room to spare: the median balance left after year 25 in the 64 successful windows is $114,445 at 18 cents, in nominal dollars. The panels in the same comparison are still running in year 26, and the model gives them no value, so the balance is not a net advantage for the index. It does show how much room the stock side had in most windows once it got through the early years.
A 1929 start at 18 cents, year by year
Start with $23,600 in the index at the beginning of 1929 and the 1,500 kWh roof at 18 cents, which saves $1,888 net in year 1 and $1,877 in year 2, falling about $11 a year. The index returns minus 8.30 percent in 1929, so the balance is $21,641 before the first withdrawal and $19,753 after it. In 1930 it returns minus 25.12 percent: $14,791, then $12,914 after the $1,877 withdrawal. In 1931, minus 43.84 percent leaves $7,253 and then $5,386.
The recovery that arrived later cannot repair a balance that small. After 1932 (minus 8.64 percent) the balance is $3,065; 1933 returns 49.98 percent but the withdrawal is $1,845, leaving $2,752; 1934 returns minus 1.19 percent and leaves $884. In 1935 the index returns 46.74 percent, which lifts $884 to $1,298 before a withdrawal of $1,824, and the balance goes to minus $526 in year 7. The solar system in the same household would have produced its 12,000 kWh in each of those years regardless of the market. The index's 25-year average from 1929 was 5.22 percent a year, and it still could not carry the withdrawals.
Rising prices: what the escalation columns show
The flat-value tables are the cautious case for solar, because stock returns are nominal and embed inflation while a flat nominal value does not. The Electric Power Annual, Table 2.4, gives the U.S. average residential price as 12.65 cents per kWh in 2015, 13.15 in 2020 and 17.23 in 2025, which is 3.14 percent a year over the ten years from 2015. That is one decade of one national average, not a forecast, and it says nothing about your utility's tariff or the export rate your panels will earn. For how rate cases and fuel costs move prices, see the EnergyBS rate guide.
The table lets the first-year value rise 2 and 3 percent a year; the 3 percent column sits just under the 3.14 percent of the last decade and the 2 percent column is a lower case. At 18 cents, 1,500 kWh per kW, the return goes from 5.7 percent flat to 7.8 percent at 2 percent a year and 8.9 percent at 3 percent a year. The 3 percent column puts the 18-cent roof above the worst stock window (5.22 percent) and above the 10th-percentile window (8.36 percent), and the 25-cent roof, at 13.3 percent, within half a point of the 90th-percentile window (13.68 percent). A reader should treat the flat and 3 percent columns as the ends of a range, not as a base case and an upside.
| First-year value of a kWh | Flat | Rising 2 percent a year | Rising 3 percent a year |
|---|---|---|---|
| 10 cents | -0.7% | 1.6% | 2.7% |
| 12 cents | 1.1% | 3.4% | 4.5% |
| 15 cents | 3.5% | 5.7% | 6.8% |
| 18 cents | 5.7% | 7.8% | 8.9% |
| 20 cents | 7.0% | 9.2% | 10.2% |
| 25 cents | 10.1% | 12.2% | 13.3% |
Escalation comes with caution where the credit for exports can fall. A utility can reset an export credit, move customers to net billing or change the hours when value is highest, so a rising retail price does not guarantee a rising value for the energy exported. The California net billing guide, the Utah export credit guide and the Iowa billing guide show three tariffs where the value of exported energy is set by a formula or reset schedule rather than by the retail price. If most of your output would be exported, use a lower column.
Tax, fees and what the comparison leaves out
The stock numbers are returns on an index, before tax and before fees. A shareholder pays tax and fund costs that the series does not. As an illustration only, the table below applies a 15 percent tax to the whole gain at sale in year 25. It lowers the median 25-year window from 10.32 to 9.67 percent and the worst from 5.22 to 4.74. The illustration understates the tax drag, because it ignores tax on dividends in the 25 years, and the 15 percent is not a statement of anyone's rate; it shows only that tax at that level takes about two-thirds of a point off the median.
| Window | Before tax | After the illustrative tax |
|---|---|---|
| Worst window | 5.22% | 4.74% |
| 10th percentile | 8.36% | 7.76% |
| Median | 10.32% | 9.67% |
| 90th percentile | 13.68% | 12.98% |
The model counts the solar savings as untaxed cash value, which treats an avoided purchase as involving no payment to the household. The solar rebate tax guide sets out how federal text treats net-metering credits and cash paid for surplus exports, with its own sources and dates. This model applies no tax to either, so a household that owes tax on export cash earns less than the tables show. If your tariff pays cash for exports or you are in a state with its own rules, the answer may differ, and a tax adviser decides it.
Other items sit outside both sides of the comparison. The solar model has no resale or residual value; the home value guide covers what sales studies measured, and the lifespan guide covers year 25 and beyond, when a system still makes 88.7 percent of its year-1 output. It has no financing cost; the loan guide computes the effective rate of a loan against the cash price, which becomes the hurdle if the money would otherwise pay down debt. It uses no state credits or rebates, since each depends on the address and the date. And it assumes the household uses or exports every kilowatt-hour at the stated value, which the sizing guide tests against the actual bill.
Risks that sit on each side
The two assets do not fail in the same way. An index return depends on markets over which the holder has no control and, as the 1929 and 2000 starts show, on when the money goes in or comes out. A rooftop return depends on production, on the tariff that sets the value of each kilowatt-hour and on the equipment and installer. The table puts each risk against the measure this page uses for it.
| Risk | S&P 500 history | Rooftop system |
|---|---|---|
| Bad early years | Worst 25-year window 5.22 percent; the 1929 start failed the withdrawal test in year 7 at 18 cents | No market sequence: the model's output falls 0.5 percent a year |
| Price or rule changes | Not applicable to an index return | The value of a kWh moves with the retail price and the export rule: the flat and rising columns bracket it |
| Output shortfall | Not applicable | 1,300 against 1,700 kWh per kW changes the 18-cent return from 4.0 to 7.3 percent |
| Equipment and company | Spread across the companies in the index | One inverter set, one installer, one roof; DOE's allowance covers replacement of components on average |
| Access to the money | Sold in days | Recovered only through lower bills, or in a sale price the model leaves out |
| Scale | Any amount | Limited by roof space and by the household's own use |
Production is the risk an owner can reduce most by checking before buying. The planning factors above are examples; a roof-specific model and the shade guide turn them into a number for one roof, and the underproduction guide covers what to do when the meter disagrees. The tariff risk is the one a buyer cannot remove, which is why the escalation table carries a flat column and two rising ones instead of a single forecast. The company risk has its own guide, what happens if the installer closes.
The two sides also differ in what the money can do afterwards. The index fund can be sold in days, and the same $23,600 can fund a car, a tuition bill or a down payment. The panels cannot be sold alone; the owner recovers the money through lower bills for as long as the household uses the power, or in a sale price that the model does not count. That illiquidity is a cost of the solar side and it does not appear in any return figure here. It weighs most for a household that expects to move within a few years.
Reading the tables for your own roof
Four steps turn the tables into a row for a particular household. First, estimate the value of a kilowatt-hour as the share used on site times the retail price plus the share exported times the export credit, using the numbers on the tariff and a recent bill. Second, take the output from a roof-specific model, in kWh per kW per year. Third, enter the all-in price per watt from the quote and find that row in the price table. Fourth, read the return column for your output and value, and compare it with the hurdle table for the benchmark you would otherwise accept, whether that is an index fund, a bond or the interest rate on a debt you would repay.
The four households in the table are illustrations built from the same inputs. Roof A, 1,300 kWh per kW at 14 cents, earns 1.2 percent and is below every benchmark in the hurdle table. Roof B, the 18-cent roof used throughout, earns 5.7 percent: above the median bond window (4.98 percent) and below the median stock window (10.32). Roof C is a less sunny roof at 30 cents, close to the August 2026 residential averages in California and Massachusetts in the EIA table, and earns 10.7 percent; the index's 25-year return was higher than that in 33 of the 74 windows. Roof D, at 1,700 kWh per kW and 25 cents, earns 12.0 percent, and the index beat it in 28. In the withdrawal test, the S&P 500 paid all 25 withdrawals in 38 of the 74 windows for Roof C and 30 for Roof D, and the bond paid in 4 and 2.
| Roof | Output and value | Year-1 net saving | Annual return | Simple payback (years) | Windows the S&P 500 paid all 25 withdrawals | Same, 10-year bond |
|---|---|---|---|---|---|---|
| A | 1,300 kWh per kW at 14 cents | $1,184 | 1.2% | 19.9 | 72 of 74 | 74 of 74 |
| B | 1,500 kWh per kW at 18 cents | $1,888 | 5.7% | 12.5 | 64 of 74 | 31 of 74 |
| C | 1,300 kWh per kW at 30 cents | $2,848 | 10.7% | 8.3 | 38 of 74 | 4 of 74 |
| D | 1,700 kWh per kW at 25 cents | $3,128 | 12.0% | 7.5 | 30 of 74 | 2 of 74 |
No row says solar is better or worse than the market. The table says what each roof returns on stated inputs, and the history says what the benchmarks returned. A household with a debt at a known rate has a third benchmark with no market risk, since paying off a 7 percent loan returns 7 percent; the cash-versus-loan guide computes that rung. And because solar savings stop at the household's own use, the choice is rarely all one or all the other: the roof is sized to the bill, and what happens to the rest of the money is a separate decision for the household.
Numbers to pull before you compare
- The tariff lines that set your value of a kilowatt-hour: the import price per kWh and the export credit, and the share of output you would use on site.
- A roof-specific annual production figure in kWh per kW, with its shading and tilt assumptions printed.
- The all-in contract price per watt DC, compared with the $2.95 benchmark line, and the scope behind any gap.
- The return you would accept elsewhere, as a rate: the debt you would repay, or the bond or index return you are comparing.
- Whether your utility is mid-way through changing its export rule, and whether the grandfathering terms would transfer if you sold the house.
Solar versus stock market questions
Is solar a better investment than the stock market?
Neither wins on every roof. At the DOE $2.95 per watt benchmark, 1,500 kWh per kW and 18 cents per kWh, the 8 kW system returns about 5.7 percent a year over 25 years with no credit, which is above the worst 25-year S&P 500 window (5.22 percent, 1929 to 1953) and below the median (10.32 percent). At 25 cents it returns 10.1 percent. The answer depends on the value of each kilowatt-hour on your tariff, your roof's output and what you would otherwise do with the money. This is history, not a forecast, and not investment advice.
What return does a residential solar system earn per year?
On the DOE benchmark system with 1,500 kWh per kW, net of the DOE running-cost allowance, no tax credit and no resale value, the 25-year return is 1.1 percent at 12 cents per kWh, 5.7 percent at 18 cents, 10.1 percent at 25 cents and 13.0 percent at 30 cents. At 1,300 or 1,700 kWh per kW the 18-cent return is 4.0 or 7.3 percent. Before the allowance the 1,500 kWh, 20-cent case is 8.5 percent.
How much must a kilowatt-hour be worth for solar to match the stock market?
On a 1,500 kWh per kW roof with a flat value, about 25.4 cents matches the median 25-year S&P 500 window from 1928 to 2025 (10.32 percent) and 17.3 cents matches the worst window (5.22 percent). If the value rises 3 percent a year, the median hurdle falls to 20.2 cents. A 1,700 kWh roof needs 22.4 cents for the median and a 1,300 kWh roof 29.3.
Is solar safer than stocks?
They carry different risks. In Damodaran's table the S&P 500 had a negative year in 26 of 98 years, and a 25-year window as low as 5.22 percent a year. A solar system's output declines slowly, about 0.5 percent a year in the DOE FEMP figure, but its return depends on the tariff, the export rule, the installer and the equipment, and the money cannot be withdrawn in pieces. This guide measures the first set of risks and brackets the second with the flat and rising columns; it does not rank them.
Does the federal solar tax credit change the comparison?
No credit is in any figure here. The IRS page for the Residential Clean Energy Credit, last reviewed July 4, 2026, says the credit is not available for any property placed in service after December 31, 2025. A state credit, rebate or certificate income would lower the price or raise the annual value and is not assumed, because each depends on the address and the date. Subtract it from the price in the price table to see its effect.
What if electricity prices rise?
A rising value helps solar and the table shows by how much. At 18 cents and 1,500 kWh per kW, a value rising 2 percent a year lifts the return from 5.7 to 7.8 percent, and 3 percent a year lifts it to 8.9 percent. The EIA residential average rose 3.14 percent a year from 2015 to 2025, one decade of a national average. Exported energy may not follow retail prices, so use a lower column if most of your output goes to the grid.
The two investments pay in different currencies. Stocks pay in dollars that can go anywhere, in amounts that cannot be known in advance; panels pay in avoided purchases, in amounts the household can estimate within a band but only up to its own use. That cap on size is why the 18-cent row is a decision about a bill and not about a portfolio. If the hurdle table says the roof clears the benchmark you would accept, the decision then reduces to the amount of solar the bill supports and the price per watt that can be documented, with the rest of the money following its own rules.
Sources and verification
- U.S. Department of Energy, Solar Photovoltaic System Cost Benchmarks (2025Q1): Table 2, PV-only: residential 8 kW DC system modeled market price $2.95 per watt DC, minimum sustainable price $2.78, O&M $34 per kW DC per year, 2024 dollars. The text excludes subsidies payable to the system owner and describes the RPV O&M elements, including the homeowner-time and property-tax and insurance sentences quoted. The page still lists 2025Q1 as the current set. $23,600 and $272 computed by SolarRoost. Checked November 4, 2026.
- U.S. Department of Energy, Federal Energy Management Program, Optimizing Solar Photovoltaic Performance for Longevity: PV module efficiency degrades at about 0.5 percent per year; PV systems have 20- to 30-year lifespans. Checked November 4, 2026.
- Aswath Damodaran, Historical Returns on Stocks, Bonds and Bills, NYU Stern (Annual returns table, 1928-2025): Annual returns for the S&P 500 (includes dividends), the 10-year Treasury bond and the 3-month Treasury bill, 98 rows, header 1928-2024 with a 2025 row, "Updated: January 5, 2026". The 2025 bond return prints 7.80 percent while the cumulative column implies 8.29 percent; 7.80 is used. Windows, percentiles and withdrawal tests computed by SolarRoost. Checked November 4, 2026.
- U.S. Energy Information Administration, Electric Power Monthly, Table 5.6.A (August 2026 data): Average residential price, August 2026: U.S. total 18.16 cents per kWh (August 2025: 17.57); lowest state Nevada 12.72, highest Hawaii 47.18. Release date October 23, 2026; next release November 24, 2026. Checked November 4, 2026.
- U.S. Energy Information Administration, Electric Power Annual, Table 2.4: Residential average price, total electric industry, nominal: 12.65 cents per kWh (2015), 13.15 (2020), 17.23 (2025). 3.14 percent a year from 2015 to 2025, computed by SolarRoost. Checked November 4, 2026.
- Internal Revenue Service, Residential Clean Energy Credit: "The credit is not available for any property placed in service after December 31, 2025." Page last reviewed or updated July 4, 2026. Checked November 4, 2026.
- SolarRoost planning assumptions: The 1,300, 1,500 and 1,700 kWh per kW production factors, the flat nominal treatment of the running-cost allowance and the 25-year horizon are this guide's assumptions, not published values for any roof. Checked November 4, 2026.
This guide is educational and is not investment, tax or financial advice. Every return is an arithmetic result from the inputs printed beside it: the DOE modeled market price, the DOE O&M allowance held flat in nominal dollars, an output decline of 0.5 percent a year, 25 years, a flat or stated rising value per kWh, no federal credit, no state incentive, no financing and no resale value. The stock, bond and bill figures are nominal, pre-tax returns from Damodaran's table for 1928 through 2025; the 74 rolling windows overlap and describe history only, and the 15 percent tax rate is an illustration. Rooftop output, tariffs and export rules differ by address and utility. Get a roof-specific production model and your own tariff before relying on any row.