A rooftop solar array on a home in Weymouth. (Photo by Senthil Balasubramanian via Wikimedia Commons/US Department of Energy)

A RECENT CommonWealth Beacon op-ed, “Before cutting solar credits, Massachusetts should do the math MIT didn’t,” argues that our research on renewables and electricity affordability is invalid because we supposedly ignored the grid-connection costs that utility-scale solar farms impose on the delivery system, while counting every cost of residential rooftop solar.

That’s an understandable worry about a certain kind of study. It isn’t a description of ours.

The op-ed takes issue with our finding that while utility-scale solar (and wind) projects are associated with lower retail rates, residential rooftop solar installations by homeowners are correlated with higher rates for the majority of state residents.

Our paper does two things the op-ed says it doesn’t. First, our primary result comes from realized residential retail prices — total utility revenue divided by total electricity sold, as reported to state and federal regulators. That number already reflects whatever utilities recover through the rate base, including the cost of upgrading a line to interconnect a solar farm. If large-scale solar farm were quietly loading costs onto the delivery system the way the op-ed describes, that would show up as a higher price coefficient on utility-scale solar in our results. It doesn’t. It shows up negative — utility-scale solar is associated with lower prices, not higher ones.

Second, and more directly on point: We didn’t stop at prices. We went and got the delivery-cost data itself. Using Federal Energy Regulatory Commission filings — the actual operating and capital expenditures utilities report by category, including transmission and distribution — we estimated how each generation technology’s share of a state’s power mix relates to delivery costs. A higher share of utility-scale solar is associated with lower transmission and distribution operating costs and lower transmission and distribution capital spending across multiple empirical models. Residential rooftop solar shows the opposite pattern on distribution costs. This is an entire section of the paper, with its own set of results tables, built for exactly the question the op-ed says we never asked.

None of this means large solar farms never require a line upgrade — of course some do, and developers typically pay for that interconnection directly, which is part of why it doesn’t show up as a rate-base cost the way the op-ed assumes. What our results say is that across 27 years and every contiguous state, the net effect of utility-scale solar on the delivery system’s costs runs in the other direction from what the op-ed claims, and we didn’t have to assume that — we measured it.

On rooftop solar, the op-ed leans heavily on a 2017 Lawrence Berkeley National Laboratory study concluding that distributed solar’s rate impact would “remain negligible for the foreseeable future.”

That study was built on national rooftop solar penetration of roughly 0.4 percent of retail sales at the time. Massachusetts today is at roughly 13 percent. Lawrence Berkeley’s own analysis didn’t claim penetration was irrelevant — it explicitly found that impacts scale with adoption and can run as high as several percent of rates in higher-penetration cases even on 2015-era data.

Our results say the same thing in a more direct way: When we split states by rooftop solar penetration, the price relationship is concentrated almost entirely in the higher-adoption half of the country, and it strengthens, not weakens, once we let each state follow its own trend.

Massachusetts sits squarely in that higher-adoption group. Citing a 2017 estimate calibrated to a fraction of a percent of national penetration as a rebuttal to what’s happening in a 13 percent-adoption state today isn’t really a rebuttal; it’s a description of a different point on the same curve we identify.

We’d also note that the literature the op-ed cites as uniformly supportive of net metering is one side of a genuinely divided empirical debate — one we discuss at length in the paper.

Researchers including Paul Joskow, and, separately, Johnson, et al. and Ansarin et al. , along with our own prior work with Ignacio Pérez-Arriaga (here and here), have found real cross-subsidization from non-solar to solar households under volumetric rate recovery.

Others, including the Lawrence Berkeley National Lab and O’Shaughnessy et al. — work that the op-ed cites — find smaller effects. Both bodies of work are legitimate; the honest reading is that the answer depends heavily on adoption levels and rate design, which is precisely our paper’s point, not a threat to it.

A simple accounting check shows why our estimated magnitude isn’t a stretch. In Massachusetts, fixed costs make up roughly 60 percent of total system costs, and about 13 percent of customers have rooftop solar and pay very little, if anything, toward those costs because net metering zeroes out their bills. When this happens, the fixed cost of running the grid doesn’t shrink; it just gets divided over a base that’s 13 percent smaller.

That alone pushes the fixed-cost portion of everyone else’s bill up by about 15 percent (13 ÷ 87), the same shift we cited in our Boston Globe op-ed on this issue, and works out to roughly a 9 percent increase in the average total bill once weighted by the fixed-cost share.

That’s arithmetic, not a regression model — and it lands in the same range as our estimated effects, which is what you’d expect if the mechanism we describe is the one actually driving the results.

To be clear about what we are and aren’t arguing. We are not, necessarily, arguing against net metering, and our study takes no position on cutting net-metering credits outright. What we found is that Massachusetts recovers most fixed grid costs through a per-kilowatt-hour charge, and that as more customers reduce their metered consumption to near zero by installing rooftop solar, the households left buying full-price power from the grid — including many lower-income households — absorb more of those fixed costs.

That’s a rate-design problem, and it has rate-design solutions: shifting more fixed-cost recovery into fixed charges, funding public-policy programs through general revenue rather than electric bills, and, at the center of it, paying solar exports what they’re actually worth to the grid.

Ultimately, rooftop solar should be compensated for the savings it provides the system — the avoided generation, the deferred upgrades, the peak demand it shaves — no more, and no less. This is exactly why we recommended that the Massachusetts DPU conduct a value-of-solar study. Getting that number right, rather than defaulting to the full retail rate, is the actual reform this debate should be about.

New Hampshire has already moved partway there. We said as much in our Globe piece, and we say it again here: The goal is a grid that’s both decarbonized and affordable, and getting there requires an accurate account of where the costs and benefits actually land — which is what our data, not the op-ed’s assumptions about our data, provide.

Fischer Espiritu Argosino is a former graduate student in MIT’s Technology and Policy Program and a research assistant at the MIT Center for Energy and Environmental Policy Research. Christopher Knittel is the George P. Shultz Professor and associate dean for climate and sustainability at the MIT Sloan School of Management and director of CEEPR.