In 2017, researchers published an unusual kind of traffic study. A major Singapore taxi operator ran two colors of cabs — yellow and blue — inherited when companies merged, with otherwise similar vehicles, drivers, and city. Comparing years of accident records, the yellow taxis had measurably fewer accidents: about nine percent fewer. The study's explanation was not driver skill or vehicle quality but visibility — yellow cabs stand out against asphalt, concrete, and the blue-gray backgrounds of a street, so other drivers, cyclists, and pedestrians notice them a fraction of a second sooner, and fractions of a second are where many collisions are decided. It is the same logic by which school bus yellow was engineered to be noticed, except here the outcome was measured in crash statistics rather than argued from paint chips.
A natural experiment on wheels
What made the data persuasive was what it controlled by accident of history. Both fleets worked the same city, the same streets, and the same company's maintenance and dispatch, over thousands of vehicles and multiple years. The researchers — who could compare the two colors at scale no laboratory could arrange — found the yellow advantage held after accounting for mileage, weather, and lighting, and their analysis suggested the effect was larger in darker hours, when the contrast between a bright yellow cab and a dim street matters most. A companion point from the same analysis: blue, the comparison color, is one of the worst choices against asphalt and evening skies, because it matches them.
Nine percent is not a magic number, and the authors were careful about what it means — the estimate applied to that fleet, in that city, over those years. But as evidence that color does safety work, it is about as clean as street data gets.
Estimating the practical value required one more step the researchers included: combining the accident gap with the cost of repainting. Switching the blue fleet to yellow, their back-of-envelope reasoning went, would prevent a meaningful number of collisions per year at the price of a paint job per vehicle — one of the rare cases where a color decision can be run through a cost-benefit calculation.
What makes yellow pop
The mechanism is figure-ground separation, the visual system's first job in any cluttered scene. Detecting an object quickly depends largely on luminance contrast — a brightness difference between the thing and what surrounds it — which is the same principle that makes text readable or invisible depending on its contrast with the background, the reason accessibility standards specify minimum contrast ratios for anything a reader must find. Yellow combines two advantages on a street: it is high in lightness, reflecting much of the light that hits it, and it is rare. Asphalt is dark and low-saturation; concrete and snow are neutral; the sky is blue or gray. A saturated bright yellow is a statistical outlier against nearly all of them, so it pops out rather than requiring search.
Pop-out matters because attention is the bottleneck. A stimulus that captures attention without effort — the way a small red badge was engineered to interrupt you — buys reaction time that a camouflaged stimulus spends on being found. In traffic, that reaction time is braking distance.
Blue taxis illustrate the inverse: high chroma but low contrast against dark pavement at dusk, and spectral overlap with both sky and shadow. The Singapore comparison was, in effect, a test of contrast against the actual backgrounds a city provides, and yellow won it.
The limits of a lucky color
None of this makes yellow universally superior, and the caveats are worth knowing. Contrast is relational: yellow that pops against asphalt can vanish against a field of rapeseed in flower, and at night what matters most is retroreflection and street lighting, not hue alone. City regulation often fixes taxi colors for identification rather than safety, mixing motives. And driver behavior swamps paint in the statistics of any individual crash — the Singapore study aggregated thousands of trips precisely because the per-trip effect is small. What the study established is narrower and more interesting: all else being equal, the color of a vehicle is not cosmetic. It is a passive device, working every hour the cab is on the street, asking nothing of the driver.
That framing — color as equipment — is the useful takeaway. Taxi operators choosing livery, cities specifying fleet colors, and anyone picking a cycling jacket or a child's umbrella are making a visibility decision, and the Singapore data gives it a number to argue with. The same logic runs through other safety palettes without needing the statistics to transfer exactly: high-visibility vests put the principle on workers' backs, and school buses and crossing guards put it at the curb. What carries over is not "yellow is safest" but the method — ask what the backgrounds are, where the observer's attention is, and what contrast the color will actually have when it matters.
If you want to feel how much detection speed depends on practice and attention rather than raw brightness, Speed Color Match times exactly that skill: rapid, confident identification of a color against its background, the pedestrian's half of the taxi's problem.
The yellow cab, it turns out, was a safety technology disguised as a brand. The paint job did not just say "taxi" — it bought every other road user a few hundred milliseconds to notice, and the accident ledger recorded the discount.