(Bloomberg Opinion) — As more and more of the world’s population shifts into cities, traffic congestion is becoming an ever-larger problem. The average American now loses around 100 hours a year sitting in traffic. Globally, congestion slows driving speeds, increasing emissions of carbon dioxide — more than 20% of which now comes from traffic.

Some dream that self-driving cars may solve the problem by smoothing out people’s natural and often disruptive driving habits, yet self-driving cars are arriving much more slowly than enthusiasts expected. Ride-hailing services might also help by reducing car ownership, but a new study shows that in cities where Uber and Lyft have been introduced, traffic delays have gone up, not down.

One obvious idea to decrease congestion is better public transportation. But experts have been skeptical of how much this can help. Economists Gilles Duranton and Matthew A. Turner argued nearly a decade ago that luring some drivers off the roads and onto trains and buses leaves less congested roadways, which then attract other drivers to those same roads. It’s similar to what happens when you try to reduce congestion by building more roads: When you make more room for cars and trucks, you get more cars and trucks.

But new research based on statistical patterns in traffic demand and the availability of public transportation flies in the face of this theory. A good measure of a city’s traffic burden is the fraction of the population that chooses to drive to work rather than use public transit. Using a simple conceptual model, physicists Vincent Verbavatz and Marc Barthelemy of the Institute of Theoretical Physics in Saclay, France, posited that easier access to public transportation can tip people away from driving.

That prediction turned out to be quite accurate for 25 large metropolitan areas in Europe, the U.S., Asia and Australia. In these cities, the fraction of people driving to work decreased in direct proportion to how easy it is to access public transportation — specifically, which fraction of the population lives within 1 kilometer of a transit station.

Why did no one discover this before? For one thing, the data didn’t exist. The research used sources such as TomTom navigation data, academic studies on access to transportation in many nations, and average driving speeds estimated from Google Maps. But more than that, Barthelemy told me by email, patterns can only be discovered if someone thinks to look for them. Their simple model suggested an interesting pattern to look for.

Their results lend credence to the idea that making public transportation more accessible can draw people away from driving, thereby reducing traffic. It also counters a fixed idea in transportation policy research that population density is the most crucial determinant of traffic patterns, with more densely populated cities being more efficient. That turns out to be wrong.

Verbavatz and Barthelemy also used their results to estimate how carbon dioxide emissions vary between cities and found that — in another prediction backed up by the empirical data — emissions rise in direct proportion to the diameter, or linear size, of a city, as well as with the proportion of people who lack good access to public transport. Compact cities emit less carbon dioxide per person, but not because of their population density. And better access to public transport always helps.

Of course, making access easier isn’t just about getting people closer to transit stations. “Easier” might mean bridging the final distance between user and transport system with convenient bus service, for example. In this sense, one of the most exciting developments in public transport might be the rise of fleets of Uber-like buses able to move people over short, flexible routes. One example is the DC MicroTransit system, currently running for free in Washington. The aim is to use mobile technology to make the mass transit system easier for everyone to access, but to do it using large vehicles so as not to tax the roadways with even more traffic.

Some experts envision the emergence of a dense microtransit mesh of public and private shuttles bringing access to sprawling public transit systems to almost everyone’s doorstep. Logically, as Verbavatz and Barthelemy’s modeling demonstrates, this seems like a sound way to reduce road traffic. The technology that drives Uber and Lyft might reduce traffic yet, just not in the way those companies initially thought.