Capitalization of urban visual morphology: a world-model perspective from the housing market of the Taiwan region
This study analyzes 544,000 housing transactions in the Taiwan region to demonstrate that urban visual morphology significantly capitalizes into property prices with distinct, income-dependent patterns that exacerbate spatial inequality, while revealing that current visual conditions drive valuations more than anticipated future changes.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
People have always chosen where to live based on what they see. A street lined with old, crumbling buildings feels different from one with fresh paint and bright storefronts; a neighborhood filled with trees and water looks distinct from a dense grid of concrete. For decades, economists have suspected that these visual differences actually change how much people are willing to pay for a home. They use a method called hedonic pricing to break down the cost of a house into its individual parts, figuring out how much value a specific feature, like a view or a quiet street, adds to the final price tag. While researchers have long studied how parks or schools affect home values, the sheer scale of modern cities makes it hard to measure the overall "look" of a neighborhood in a consistent way. Until recently, capturing the visual texture of an entire city required expensive, hand-drawn maps or unreliable surveys. Now, with powerful computers and satellite cameras, scientists can measure the visual shape of cities with a precision that was previously impossible, allowing them to ask a simple but profound question: does the visual appearance of a neighborhood actually show up in the price of a house, and does that value change depending on who is buying?
A team of researchers set out to answer this by looking at the housing market across the Taiwan region, analyzing a massive dataset of nearly 544,000 real estate transactions recorded between 2012 and 2026. To understand the visual environment of each home, they did not rely on human observers. Instead, they used satellite images taken from space to create a detailed visual map of the area surrounding every single house sold. For each property, they examined a half-kilometer circle around it, calculating eight specific visual features. They measured how much of the ground was covered by green plants or water, how much of the built environment appeared bright or dark, and how complex the texture of the buildings was. They also looked at how much the brightness varied from one spot to another, which serves as a measure of visual disorder. By matching these visual measurements to the actual sale prices of the homes, while carefully controlling for the age and size of the buildings and the specific year of the sale, they could isolate exactly how much buyers were paying for the visual look of their neighborhood.
The results revealed a clear and surprising pattern in how people value what they see. The researchers found that a neighborhood with a complex, intricate texture of buildings—where the visual pattern is busy and detailed—commands a significant premium, meaning people are willing to pay more for it. In contrast, neighborhoods with a high share of green pixels, water pixels, or areas that are very bright or very dark were actually discounted, selling for less than similar homes in areas without these features. This negative value for green and water at this specific local scale was not because people dislike nature, but because in the dense urban cores of this region, these features often mark older, lower-density areas that are already less expensive for other reasons. The study showed that the market rewards visual complexity and penalizes visual disorder, but the story changes when you look at how these values have shifted over time. Over the last decade, the market has become even more critical of dark and green built forms, while the value of bright, compact built forms has risen.
Perhaps the most striking discovery was that these visual preferences are not shared equally by everyone. The value of a neighborhood's appearance depends heavily on the income of the people living there. In high-income, wealthy neighborhoods, buyers place a high premium on brightness and are willing to pay extra for it, while they strongly dislike complex, textured environments. However, in low-income communities, the opposite is true. For residents in these areas, a brighter built environment is seen as a tangible sign of improvement and carries a positive value, whereas the complex textures that wealthy buyers dislike are not penalized in the same way. This creates a situation where the visual environment acts as a mirror of social inequality: the features that wealthy communities pay for are different from the features that poorer communities value, and the financial benefits of a "good" visual environment tend to flow to those who are already better off.
To ensure these findings were not just a fluke of their specific measurements, the researchers tested their results against a massive, advanced artificial intelligence model designed to understand satellite imagery. This model, which had been trained on millions of images from around the world, could see the visual world in thousands of different ways. The researchers found that their simple, eight-feature description of the neighborhood was almost as good at predicting housing prices as the complex, high-tech model. The advanced model added almost no new information, suggesting that the human-interpretable features they measured—like brightness and texture complexity—capture the essential visual information that the housing market cares about. This also meant that their results were robust and not dependent on a "black box" algorithm that no one could understand.
The team also explored whether people were paying for the future, betting on neighborhoods that they thought would get better looking over time. They treated each neighborhood as a small system that changes over the years, tracking how its visual appearance evolved. They found that the housing market does not seem to anticipate future changes. Instead, prices reflect the visible present. If a neighborhood is on a trajectory to become brighter or more complex in the future, the market does not pay extra for that potential today. In fact, in some recent cases, neighborhoods that were improving visually were priced slightly lower, suggesting that buyers are focused entirely on what they can see right now, rather than what the neighborhood might become. The visual future is so similar to the visual present that there is very little room for speculation.
Finally, the researchers used their findings to run a simulation of what would happen if city planners could change the visual appearance of neighborhoods. They calculated the potential financial impact of reducing the amount of dark, built-up areas or increasing the brightness in the darkest parts of the city. Their calculations suggested that if these visual interventions were made, the total value of homes across the region could increase by roughly 350 billion New Taiwan dollars. This simulation highlights that the visual environment is not just a matter of aesthetics; it is a tangible economic asset. However, the researchers cautioned that these numbers are based on a snapshot of current preferences and do not account for how the supply of housing might change in response. The study concludes that the way we see our cities is deeply woven into the economy, shaping who can afford to live where and reinforcing the spatial divides that already exist in society.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.