Rail trails and property values

linear regression
nonparametric regression
Rail trails are a great recreation opportunity, but do homes near them rise in value? A retrospective observational study of house values.
Author

Alex Reinhart

Published

July 27, 2022

Data files
Data year

2014

Motivation

In the late 1800s and early 1900s, numerous rail lines were built across the United States and connected most towns and cities for passenger and cargo service. But as cars became dominant and the Interstate Highway System made long-distance travel by car easier, rail traffic declined, and many rail lines were shut down or abandoned. Some lines were kept by the railroads for future use, while others were sold off.

Beginning in the 1980s, some unused rail lines began to be converted to rail trails: walking and biking trails along the route of an old rail line. Rail trails tend to be long continuous trails, often paved, and have gentle slopes at most (because trains cannot climb steep hills). This makes them accessible and easy to bike along.

Some have suggested that rail trails may be an attractive feature for people buying homes, who might be willing to pay more for a home close to a trail where they could walk or bike for fun or for their commute.

This data was collected from houses sold in Northampton, Massachusetts in 2007. A new rail trail was opened in Northampton in 1984, offering an opportunity to compare the values of homes near the trail and farther from the trail. See the References below for more detail on the data collection and the history of rail trails in the area, which provides useful context.

Data

The data includes 104 homes in the Northampton, Massachusetts area that were sold in 2007. Attributes of the homes were collected (by Hartenian and Horton; see References below) from several sources:

  • Values came from Zillow’s value estimator, which is calculated by Zillow based on sale prices, prices of nearby similar homes, and features of the home
  • Distances from the homes to the rail trails were obtained using Google Maps
  • Bike and walking scores came from Walkscore.com
  • House attributes (such as size and number of rooms) came from the Multiple Listing Service (MLS), which is used by real estate agents to list houses for sale. The values obtained were from 2007

The data only includes homes with under 0.56 acres of property, and excludes two homes that increased in value more than $500,000 between 1998 and 2007 due to major renovations, which were regarded as outliers.

Data preview

rail-trails.csv

Variable descriptions

Variable Description
housenum A unique number for each house
housenum.1 Accidental duplicate of the housenum column
adj1998 Zillow’s estimated value for the home in 1998, in thousands of 2014 dollars
adj2007 Zillow’s estimated value for the home in 2007, in thousands of 2014 dollars
adj2011 Zillow’s estimated value for the home in 2011, in thousands of 2014 dollars
price1998 Zillow’s estimated value for the home in 1998, in thousands of dollars
price2007 Zillow’s estimated value for the home in 2007, in thousands of dollars
price2011 Zillow’s estimated value for the home in 2011, in thousands of dollars
price2014 Zillow’s estimated value for the home in 2014, in thousands of dollars
diff2014 Difference in price between 2014 estimate and 1998 estimate, in thousands of 2014 dollars
pctchange Percentage change from adjusted 1998 value to 2014 value (0 means no change)
distance Distance (miles) to the nearest entry to the rail trail network
distgroup Distance to the rail trail network, categorized as “Closer” (less than or equal to half a mile) or “Farther Away” (more than half a mile)
acre Number of acres of property
acregroup Indicates if the number of acres is smaller or larger than 1/4 acre
bedgroup How many bedrooms the home has, categorized to 1-2, 3, or 4+
bedrooms How many bedrooms the home has
bikescore Bike friendliness of the area, estimated by WalkScore.com. 0-100 scale, where 100 indicates high bike-frinedliness, such as flat terrain and good bike lanes.
walkscore Walkability of the area, estimated by WalkScore.com. 0-100 scale, where 100 indicates high walkability, so most daily tasks can be done without a car
garage_spaces Number of garage parking spaces (0-4)
garagegroup Does the house have garage parking? (yes/no)
latitude House’s latitude
longitude House’s longitude
no_full_baths Number of full bathrooms (i.e. bathrooms with a shower or bathtub)
no_half_baths Number of half baths (i.e. bathrooms with no shower or bathtub)
no_rooms Number of rooms
sfgroup Whether the home has more or less than 1500 square feet of interior space
squarefeet Square footage of the home’s interior finished space (in thousands of square feet)
streetname Name of the street the house is on
streetno House number on the street
zip ZIP code of the house (leading 0 omitted). 1060 is Northampton, MA; 1062 is Florence, MA.

Questions

  1. This dataset only includes homes sold in 2007, rather than all homes in Northampton or a random sample of homes. Comment on what biases this might induce and how this might affect our analysis results.
  2. Conduct a simple analysis using 1998 prices and 2014 prices, and the distance from the homes to the trails. Did homes closer to trails appear to appreciate more?
  3. Houses close to the trail may differ systematically from those farther away. Conduct a more detailed analysis that controls for home features, such as the size and number of bedrooms, and comment on your conclusions.

References

Ella Hartenian and Nicholas J. Horton (2015). Rail Trails and Property Values: Is There an Association? Journal of Statistics Education 23:2, https://doi.org/10.1080/10691898.2015.11889735