The website. (Andrew Harnik/Associated Press)

IT IS often said that good policy makes for good politics. And we still want to believe that, despite the history of the “Cadillac tax” in the Affordable Care Act.

Current law excludes employer-paid health insurance from taxation, which is how nearly half the public gets coverage. This subsidy, which costs $250 billion per year, according to the Congressional Budget Office, nevertheless promotes overutilization of health care, thus driving cost inflation. It causes “job lock” by linking work and insurance and redistributes income upward because tax breaks are worth more to higher income brackets. Nothing would improve U.S. health care more than to repeal the exclusion and use the savings to fund a more rational system. That’s politically impossible, however. So the ACA, passed in 2010, incorporated a second-best solution: a 40 percent levy on relatively plush “Cadillac” employer-paid plans (above $10,200 for individuals; $27,500 for families), effective 2018.

That delay was forced by opponents of the idea, who included not only the usual anti-tax business lobbies and Republicans but also labor unions, who feared losing their ability to negotiate ever-improving health benefits. And these interests kept up their attacks on the tax after the ACA’s passage, even though the mere anticipation of it was stimulating cost-saving innovation by companies and insurers. The opponents achieved a major victory by tacking an additional two-year implementation delay, until 2020, onto must-pass tax legislation that President Obama signed in December.

Now, the provision is hanging by a political thread. The GOP House is on record against the entire ACA; a bipartisan, filibuster-proof majority of the Senate is on record favoring repeal of the Cadillac tax; and all major presidential candidates are hostile to the provision as well.

The latest development is a proposal in Mr. Obama’s newly released fiscal 2017 budget that would further adjust the Cadillac tax by taking account of regional differences in health-care costs. Basically, in any state where the average premium for “gold” coverage on the state’s individual health insurance marketplace exceeds the Cadillac-tax threshold, the tax would kick in only on employer-paid plans when they hit that average gold premium. Economically, the idea is to ease unintended burdens on companies located in high-cost areas; politically, the idea is to make the representatives and senators from those areas, and perhaps others, more amenable to letting the tax go into effect in 2020 after all.

The downside is to sacrifice both revenue and cost-control in the short run — the proposal arguably reduces downward pressure on costs precisely where they are most in need of it. Still, some of that extra cost is due to demography and other structural causes for which it may be appropriate to make allowances. And, once in effect, the tax would still have a steadily growing impact due to its inflation-adjustment terms. It is depressing indeed to see the administration forced to fight this rear-guard action for a policy economists widely recognize as one of its smartest. But it will be worth it if what emerges is a more politically viable version — and a reason to hope that Washington will not always be the place good ideas go to die.