I create a time series of weekly ratios of Google searches, in the US, on buying and selling in the Real Estate Category of Google Trends. I call this ratio the Google US Housing Market BUSE Index or simply the BUSE index. It expresses the number of "buy"-searches for each "sell"-search which, by means of certain regularity assumptions on the distribution of Internet users, I think is a good proxy of the number of prospective home buyers for each prospective home seller in the pool of prospective housing market participants. I show this ratio to have several unique, desirable properties which make it useful for understanding and nowcasting the US housing market. Firstly it has a significant correlation with the US national S&P/Case-Shiller Home Price Index. Since the latter is monthly and published as a three-month moving average with a two month lag and the Google Trends data is weekly we can have a short term nowcasting of housing prices in the US. In the seasonal variations of this ratio the BUSE index recaptures traces of prospect theory whose applicability in the housing market has been well documented. I show how these Google data can be used to create a consistent narrative of the post bubble burst dynamics in the US housing market and propose the BUSE index as an instrument for monitoring housing market conditions.
computational social science data science behaviour complexity S&P/Case-Shiller Home Price Google Search Google Trends housing market nowcasting complex systems