Forecasting house prices to optimize investments for a real estate client

Project Details

Term: 

2019

Students: 

Shrisha Kashyap, Harshal Chaudhari, Prateek Agarwal, Shashank Gupta, Anuj Bairoliya, Sriram Valavala

University: 

ISB

Presentation: 

Report: 

We have forecasted time series of House Price Index to help a Real Estate company invest in a city. Our analysis recommends investing in Dallas to get 38% gain in 2 years.

 

  • Purpose: to recommend a location(/s) for our client, a real estate developer in the United States, such that the return on investment for the housing venture is maximum
  • Selection of cities: six cities with the highest GDP growth are selected – New York, San Francisco, Chicago, Washington, Dallas and Los Angeles
  • Data Source: House Price Index (HPI). Federal Housing Finance Agency (FHFA) publishes the index every quarter and is defined as “a weighted, repeat-sales index -average price changes in repeat sales or refinancing on the same properties (1)”. Base quarter Q1 1991 = 100 and the values are comparable over periods and across locations
  • Data Processing: The data was pre-processed to remove the values from Q1 1991 to Q4 2007 so that the event - 2007 housing market crash - is avoided. The forecast horizon is two years (Q3 2016 to Q2 2018).
  • Forecasting method shortlisted based on similarity of % gains on validation: We shortlisted the best forecasting method based on similarity of the % gains over the validation period for model and actual, along with RMSE & MAPE for validation period.
  • Recommendations: plot depicts the time series for HPI along with the forecast values (shown in dotted lines). Based on the results of our analyses, we recommend choosing Dallas, TX because its HPI is estimated to grow the highest (38%) in the next two years.
  • Risks: The company should diversify the risk by investing in more than 1 location and investigate tax and bank considerations.

Application Area: