Can Shared Rides Fix Our Cities?

The Problem

Ride-hailing was supposed to fix traffic. It made it worse.

When Uber and Lyft arrived, the promise was seductive: fewer privately owned cars, smarter routing, less congestion. The reality has been almost the opposite. Multiple studies across San Francisco, Chicago, and other cities confirm that ride-hailing increases vehicle kilometers traveled (VKT) — the total distance all vehicles drive — because these services pull passengers away from buses and metros, and add enormous numbers of "deadhead" miles while drivers cruise looking for their next fare.

But here's the question that has gone largely unanswered: what about apps that don't send a car exclusively for you — apps designed from the ground up for shared trips? Could a platform that deliberately puts multiple strangers in the same van, travelling the same corridor, change the equation?

"There is little evidence if smartphone apps that target shared rides have any influence on reducing traffic levels."

That gap in the literature is precisely what this paper targets. The researchers studied Jetty, a Mexican start-up operating in Mexico City — one of the world's largest and most congested metropolitan areas — that lets commuters book a seat in a shared car, van, or bus through a smartphone app.

The Setting

Mexico City: a megacity with a mobility crisis

The Mexico City metro area is home to 21 million people. On a typical weekday, 34.5 million trips are made. The modal split reflects enormous complexity: 40% public transit, 32% walking, 19% private car, 5% taxi or app-based ride.

Public transit includes a 12-line Metro, BRT (Metrobus), and millions of lightly regulated colectivos — minibuses, microbuses, and vans that account for over 11 million daily trips. These colectivos cover the gaps that formal transit can't, but they vary wildly in quality, safety, and reliability.

Into this environment came Jetty (launched 2017). Unlike ride-hailing, Jetty doesn't own vehicles — it connects commuters with licensed operators running fixed routes, primarily serving two major job clusters: Santa Fe (a major business district) and Polanco (an upmarket commercial area). Vehicles pick up passengers at designated stops on a published schedule. Think of it as a premium, app-bookable minibus — more expensive than a colectivo, but far cheaper than a private taxi.

Who Uses It

The riders: wealthier, more educated, and mostly car owners

The research team conducted a large-scale survey of 2,484 Jetty users. The demographic picture is striking and crucial to interpreting the results.

Top reasons riders choose Jetty (% who mentioned it)

Booking a seat

70%

Security/theft

68%

Travel time

66%

Reliability

53%

Access/egress

42%

Vehicle quality

36%

Fare

36%

Anti-harassment (women: 27%)

16%

The most notable finding: 80% of Jetty users have at least one car at home — nearly double Mexico City's average of 41%. This is crucial. If these commuters would otherwise be driving solo, replacing those trips with a shared van is a genuine win for traffic. The platform is targeting exactly the right demographic: people with cars who are choosing not to drive them.

The gender gap in the harassment statistic deserves attention too: 27% of women cited protection from sexual harassment as a reason for using Jetty, versus just 4% of men. This points to a quality-of-life dimension of transport infrastructure that aggregate VKT numbers simply cannot capture.

The Core Finding

Vehicle type is everything — and the results are surprising

The central question: does Jetty add or remove kilometers from Mexico City's roads? The answer depends almost entirely on which vehicle you're riding in. Click each vehicle below to explore why.

🚗 Shared Car +7–10 km VKT per passenger

🚐 Shared Van −0.2 to −1.1 km VKT per passenger

🚌 Shared Bus +0.4–1.1 km VKT per passenger

Why do shared cars increase traffic? Even though these are "shared" (3–6 seats), the average demand per run is only around 2 passengers. The car adds nearly as many vehicle-kilometers as a solo driver would, but with only marginally more riders. The VKT ratio — vehicle-km per passenger-km — is 3.5× for cars vs. below 1× for large buses. Cars also predominantly replace trips that were already made in low-occupancy vehicles (taxis, ridesourcing), so the network barely benefits.

Why do vans actually reduce traffic? Two factors combine favorably here. First, vans carry 7–10 passengers on average, spreading the vehicle-kilometers across many riders. Second, and more importantly, van users disproportionately replace car trips: 15% would have driven themselves if Jetty weren't available, versus only 10% of bus users. Taking a car off the road and replacing it with a shared van carrying 9 people is a clear win. The VKT savings are modest but real — and importantly, they survive even with 10–20% empty-kilometer assumptions.

Why do buses increase VKT despite high capacity? This is the paper's most counterintuitive result. Buses carry up to 18 passengers per run on average and have excellent VKT-per-passenger ratios. But their riders mostly came from public transit in the first place — 23% replaced a Metro + Microbus combination. They weren't driving; they were already in shared modes. Adding a premium bus to the network doesn't remove cars — it just offers a nicer version of a trip that was already efficient. The extra vehicle-kilometers of the Jetty bus outweigh the marginal improvement in their replaced trips.

The key isn't vehicle size. It's who switches — and what they were doing before.

Interactive Explorer

Run your own scenario

The overall VKT effect of a shared-mobility platform depends on several variables the researchers tested through sensitivity analysis. Try changing the assumptions below — just as the researchers did.

VKT Impact Estimator

Based on Tirachini et al. (2020) — Scenario A family

Vehicle Mix

Supply Adjustment

Empty kilometers: 10%

0%10%20%30%

—

Select options above

Adjust the controls to see how VKT changes under different assumptions. This directly mirrors the sensitivity analysis in the paper.

Key Findings

Five things this research tells us

01
Shared-mobility apps can reduce traffic — under the right conditions Unlike ride-hailing, a well-designed shared platform operating medium-size vans can demonstrably reduce VKT, especially when it draws users away from private cars. This had not been shown empirically before.
02
Modal substitution matters more than vehicle capacity The study shows buses have worse VKT outcomes than vans — not because buses are inefficient, but because bus riders mostly came from public transit. The crucial question is: what were people doing before?
03
Empty kilometers are a critical variable Vehicles deadheading between routes can quickly erase all VKT savings. Efficient depot placement and scheduling optimization aren't just operational details — they determine whether a shared platform is sustainable at all.
04
Quality attributes drive adoption as much as price Jetty costs far more than colectivos, yet users rank guaranteed seating and security higher than fare. This suggests that improving transit quality — not just expanding capacity — could retain and attract riders.
05
Equity must be part of the equation Jetty's users are disproportionately wealthy and educated. Any integration of premium shared-mobility into public transport systems will need explicit subsidy mechanisms to avoid entrenching a two-tier system.

Methodology

How was this measured?

The paper uses a carefully designed counterfactual approach. Here are the core methodological questions answered.

The key survey question was: "If Jetty was not available, how would you have made your latest Jetty trip?" Respondents could choose up to three alternative modes, indicating a multi-modal chain. Researchers then used Google Maps routing APIs to calculate the distance of those alternative trips, compared to the distance actually traveled by the Jetty vehicle. VKT for Jetty vehicles was calculated from actual trip data provided by the company, adjusted for passenger car equivalency (PCE) factors — larger vehicles count as more "car equivalents" in traffic models.

The researchers model three assumptions about how other modes respond to Jetty drawing away riders. Long-term (all modes adjust): buses and metros reduce their services proportionally to lost ridership, so the system-wide VKT decreases. Medium-term (low-occupancy modes only): only private cars, taxis, and ridesourcing reduce their driving; public transit keeps running the same schedules. Short-term (only cars): only private car drivers change behavior; everything else continues as normal. The realistic scenarios sit somewhere between medium and long-term.

14,093 users were invited; 3,091 responded (~23% response rate); 2,484 completed the survey; 2,169 were usable after quality checks. The sample broadly matches Mexico City demographics for age, gender, and household size. However, Jetty's user base — and therefore the sample — over-represents higher-income, highly-educated car owners relative to the general population. This is acknowledged as a limitation, but also reflects Jetty's actual market: the platform is genuinely targeting people with cars.

Several key inputs are genuinely unknown — for example, the average occupancy of Mexico City colectivos, or the exact percentage of empty kilometers Jetty vehicles travel between routes. Rather than picking arbitrary defaults, the researchers performed sensitivity analyses: they ran the calculation with high, medium, and low occupancy assumptions, and with 0–30% empty-kilometer assumptions. The result is a range of outcomes that honestly reflects this uncertainty. The paper's contribution is to identify the conditions under which VKT savings occur, not to claim a single definitive number.

Can shared rides actually cut city traffic?