Note: Dependent variable is natural log of transacted house prices. Housing structure characteristics include size, number of rooms, fireplace, new build, energy efficiency, flat type, and tenure. Location controls include distance from various nature features (river, lakes, docks, bridges, embankments, reservoirs), transportation nodes (rail, rapid lines), and town center. Land cover controls include dummy variables for land uses around the postcode that include urban land, suburban, arable land, grassland, rock land, woodland. Employment controls include share of employment in various sectors (e.g., manufacturing, wholesale, retail) collected at the LSOA of postcode. Neighborhood controls include unemployment rate, population density, population size, percentage of residents who are illiterate, non-European, nonwhite, and percentage of households with lone parents, without cars, who are social renters, who are homeowners. All these neighborhood controls are collected at OA level. See Appendix Table A1 for more details on the variables included. Absolute price change is computed by multiplying the average transacted prices with the estimated premium for properties within 100 meters from the nearest canal. Standard errors are clustered at LSOA level in columns (1), (3), and (4) and clustered at MSOA level in column (2).

↵*, **, *** denote significance level at 10%, 5%, and 1%, respectively.