A new study quantifying the impact of curb appeal found that the attractiveness of a home can boost its value by 7% or more
By Adam Bonislawski - The Wall Street Journal Jan. 23, 2020
Want to get top dollar for your house? Mow the lawn.
In a recent study in the Journal of Real Estate Finance and Economics, researchers at the
University of Alabama and the University of Texas at Arlington used deep learning and Google Street View to determine just how much curb appeal contributes to a home’s value.
Analyzing Google Street View photos and sales data from 88,980 properties in the greater
Denver area, the researchers determined that on average a home with excellent curb appeal sold for 7% more than a similar house in the same neighborhood with poor curb appeal. In slow real-estate markets (when buyers can afford to be choosier), that premium rose to as high as 14%.
That 7% figure also factors in the state of the home across the street, which accounted for
roughly a third of the overall premium. So, pick your neighbors carefully. Granted, the notion that people prefer a nice yard isn’t exactly surprising, but while the findings aren’t game changing, the way the researchers arrived at them could be, says Sriram Villupuram, an associate professor at UT Arlington and author of the study along with Erik Johnson and Alan Tidwell from the University of Alabama.
Everyone knows that curb appeal affects a property’s value, but quantifying that impact presents a challenge, Dr. Villupuram notes. Data on things like bedroom count, square footage and lot size are easily obtainable for most homes and are commonly incorporated into appraisers’ models. Curb appeal, on the other hand, is more difficult and labor intensive to account for, with assessments often requiring in-person visits. “It’s observable, but not quantifiable,” Dr. Villupuram says. “And we have tried to change that with this paper.”
To get at the question of curb appeal, the researchers manually scored a set of properties, grading 400 images on a scale of 1 to 4, with 1 indicating the lowest appeal and 4 the highest. Low-appeal properties had blemishes like broken pavement and overgrown grass, while high-appeal properties were characterized by features like well-kept lawns and nice landscaping.
The researchers then used these scored images to train their deep-learning algorithm to assign curb-appeal values. Lastly, they used the algorithm to grade the photos in the larger data set. Comparing sales prices of homes with good curb appeal to those with bad curb appeal (and controlling for factors like neighborhood, time of sale, and house size and features) they arrived at their finding of a premium of 7%.
The goal, Dr. Villupuram says, is to automate assessments of curb appeal, which could make it easier for large investors, banks and institutions like Fannie Mae to include these assessments in their property appraisals. The algorithm isn’t perfect. In the study, it assigned curb appeal correctly about 66% of the time when compared with manually scored photos. Using larger and more geographically diverse data sets to train the algorithm should boost its accuracy, Dr. Villupuram notes. “There’s definitely room for improvement,” he says.
Speaking of which—you rake those leaves up yet?
