Traffic is awful. It causes pollution. It stressed people out. It costs cities and drivers billions of dollars. And if you’ve ever sat in a car, inching along a packed highway, you understand the toll it takes.

So, how do we fix it? According to UCLA’s Michael Manville, there are a lot of proposed solutions, but only one — congestion pricing — really works.

First, let's tackle why other solutions don't really fix the problem.

A common proposal is to build more roads, or to widen existing roads. That’s why the 405 freeway in Los Angeles has so many lanes. The thinking goes: if more cars can fit on the road at the same time, well then, traffic has to speed up. Right? Well, yes … but only for a short while. And that increased speed makes driving on the road more attractive. So, within a matter of a few weeks or months, the improved traffic has convinced more people that: hey, the roads aren’t so bad. And once that happens, they’re terrible all over again.

Well then, what about public transportation? Manville believes public transportation is great, and it makes sense for cities and states to invest in it. (Here’s looking at you, MTA) But he doesn’t believe it’s a cure for congestion. Think of the places in the world which have the most comprehensive public transportation systems — places like Tokyo, London, and Hong Kong. They still have terrible traffic congestion.

So what will work? Perhaps a more densely populated urban center, with affordable housing closer to where people actually live? Once again, Manville thinks that affordable housing would probably be great … but it still wouldn’t cure traffic. He pointed at cities with dense city centers, like New York or Shanghai — they still have terrible traffic.

According to Manville, there is only one real, proven solution to traffic: price our roads. What he proposes is a type of toll called a congestion charge. The toll is dynamic, with the level rising or falling based on demand for the road at a specific time. So at midnight, when few cars are on the road, it might be low, whereas at rush hour, when a bunch of people are trying to get to work, it would be higher. Unlike other tolls, it’s not designed to raise money; instead, it’s designed to make a few people decide not to drive.

"The majority of the delay is caused by the last few vehicles on the road,” a road can absorb a lot of cars and still be traffic-free. But after a certain point, each additional vehicle causes a lot of delay, Manville said.

Since traffic is caused by these last few vehicles, Manvilles thinks that “you only need a toll high enough to get 4 percent or 5 percent of the drivers off, and that could lead to something like a 15-20 percent increase in speed.”

But what about people that would have difficulty paying the toll? Would that unfairly impact them? Perhaps. But Manville says this could be fixed. He envisions using some of the revenue created by congestion pricing to subsidize the commutes of folks with lower incomes — much in the same way that some heating and electricity is subsidized.

And Manville says that there’s evidence that congestion pricing works. Singapore has used some versionof it since 1975. Cities like London and Stockholm have also tried congestion pricing.

As for whether this will actually happen in America? Manville thinks it’ll be a hard sell, especially since people are used to roads being free. ‘But I’m optimistic,” Manville says, “that as more and more places, many of them outside the United States, experiment with this, mayors in the United States will look around and say ‘there’s no reason we shouldn’t be part of this success.’”