Last month, a narrowly divided Supreme Court left it to the states to settle issues of gerrymandering. Yet, while almost everybody seems to oppose gerrymandering in theory, solutions that meet everyone’s standard of fairness are elusive. Indeed, all of the remedies proposed by courts, legislatures and academics face serious problems because they attempt to solve the problem the hard way. As in many areas of law, designing optimal regulations that are transparent and that can be applied over wide range of circumstances is frustratingly difficult.

By contrast, creating incentives for parties to align their behavior with social objectives is often much easier. Just as the general threat of legal liability gives companies incentives to make safe products without regulation of product design, deterring districting partisanship does not require before-the-fact regulation. The idea is to incentivize state legislatures or redistricting commissions to draw fair boundaries in the first place by making rule-based adjustment of election results after the fact if the results deviate from easy-to-measure objective standards.

To illustrate, consider a state with 10 congressional districts. Imagine the House popular vote in that state is 55 percent Republican and 45 percent Democrat, implying that a Republican win of 5 or 6 seats would be consistent with fair districting. But, if the Republicans win more than 6 seats or fewer than 4 seats, my approach would let one or more losing candidates from the disadvantaged party to “borrow” votes (with voter’s consent, as explained below) going to candidates from their own party in other districts if the borrowing does not change the outcome(s) of those other races.

That’s a mouthful. To make things more concrete, imagine that the Republicans won just 3 out of 10 seats in this hypothetical gerrymandered election, even though fairness would have dictated at least 5 Republican victories. To remedy that, start with the Republican candidate who received the 4th highest vote fraction (the Republican candidate who lost by the smallest margin). If she lost by, say, 1,000 votes, she could take 1,001 Republican votes cast in any other districts — provided the transfer did not change the results in the other district. Ideally, this transfer could be made contingent on voters’ consent by giving them a chance to specify on their ballots which out-of-district candidates they supported.

If there are at least 1,001 such votes across the state, then this Republican is declared the winner in this race. But the fix need not stop there. We use the same procedure for the Republican candidate who lost by the second-smallest margin. Say this candidate lost by 6,400 votes. He can borrow up to 6,401 Republican votes from other races, as long as this doesn’t change the outcome of those races. If so, this Republican candidate is also declared the winner.

The process stops with 5 Republicans and 5 Democrats declared winners, an outcome that is consistent with the state’s popular vote. If the gerrymandering yielded in more skewed results favoring Democrats, we would continue with the next closest Republican to victory until the Republican-Democrat split in seats is consistent with the state’s overall vote.

This procedure would introduce new strategic considerations in elections. First, political parties would have an incentive to turn out their supporters even in races that were not in doubt. Second, candidates might have an incentive to campaign outside their own districts to generate more votes to borrow. In this age of hyper-partisanship, however, there would probably be enough voters willing to lend their votes to like-minded candidates outside the voters’ own districts without additional campaigning. Third, parties might have an incentive to nominate candidates with cross-party appeal in order to raise their total vote share. This, by my way of thinking, would be an added bonus.

Of these three incentives, note that only the third would introduce a genuine issue of what constitutes good public policy — everybody, in theory, wants “fair” elections not subject to gerrymandering. Note, too, that if the system is working properly, vote borrowing across districts would rarely occur. After all, if partisan state legislatures knew they can’t affect the division of seats through gerrymandering, they wouldn’t have an incentive to try.

A key consequence of this approach to gerrymandering is that it doesn’t require any before-the-fact analysis by a third party (say, a by special commission appointed by a court) of what constitutes fair districting. It is enforced entirely after an election, based on easily observable measures — a reality that should especially please conservative judges who have shown reluctance to intervene in the nuts and bolts of voting even when redistricting undermines fair elections.

Doubtless there would be reflective skepticism about adding to the complexity of elections, as there have been to other proposals to more accurately capture the will of voters, such as Maine’s ranked-choice voting system. But it’s the approach the law takes in many other areas (and the approach that is as natural as breathing to economists). Instead of trying to dictate the appropriate outcome, give the parties making the decision the right incentives, and they will generate an appropriate outcome on their own.