Traditionally, hedonic pricing studies focusing on commercial real estate control for spatial effects with macro-location dummy variables, for example, at the MSA or submarket level, and predominantly rely on OLS regression. This approach ignores the importance of micro-location characteristics for predicting sales prices. Additionally, spatial dependencies in transaction prices and error terms violate OLS assumptions. We propose an alternative approach that replaces macro-location dummy variables with four location area characteristic (LAC) variables, defined as population, median home value, homeownership rate, and average land gradient within a polygon of a 20-minute driving distance around each property in our sample. We compare the location dummy and LAC approach using OLS regression, spatial autoregression (SAR), and spatial autoregression with autoregressive errors (SARAR) for a sample of hotels sold between 2015 and 2017. We find models that include LAC variables to have a superior fit to models with MSA dummy variables, irrespective of estimation method. Additionally, the inclusion of individual LAC variables allows a reduction of spatial lags and errors.