Driving forces for the US residential housing price: a predictive analysis

Purpose

The purpose of this paper is to determine the key characteristics that determine housing prices in the USA. Data analytical models capable of predicting the driving forces of housing prices can be extremely useful in the built environment and real estate decision-making processes.

Design/methodology/approach

A data set of 13,771 houses is extracted from the 2013 American Housing Survey (AHS) data and used to develop a Hedonic Pricing Method (HPM). Besides, a data set of 22 houses in the city of San Francisco, CA is extracted from Redfin real estate brokerage database and used to test and validate the model. A correlation analysis is performed and a stepwise regression model is developed. Also, the best subsets regression model is selected to be used in HPM and a semi-log HPM is proposed to reduce the problem of heteroscedasticity.

Findings

Results show that the main driving force for housing transaction price in the USA is the square footage of the unit, followed by its location, and its number of bathrooms and bedrooms. The results also show that the impact of neighborhood characteristics (such as distance to open spaces and business centers) on the housing prices is not as strong as the impact of housing unit characteristics and location characteristics.

Research limitations/implications

An important limitation of this study is the lack of detailed housing attribute variables in the AHS data set. The accuracy of the prediction model could be increased by having a greater number of information regarding neighborhood and regional characteristics. Also, considering the macro business environment such as the inflation rate, the interest rates, the supply and demand for housing, and the unemployment rates, among others could increase the accuracy of the model. The authors hope that the presented study spurs additional research into this topic for further investigation.

Practical implications

The developed framework which is capable of predicting the driving forces of housing prices and predict the market values based on those factors could be useful in the built environment and real estate decision-making processes. Researchers can also build upon the developed framework to develop more sophisticated predictive models that benefit from a more diverse set of factors.

Social implications

Finally, predictive models of housing price can help develop user-friendly interfaces and mobile applications for home buyers to better evaluate their purchase choices.

Originality/value

Identification of the key driving forces that determine housing prices on real-world data from the 2013 AHS, and development of a prediction model for housing prices based on the studied data have made the presented research original and unique.