Can Online Searches Predict Price Hikes In China Property Markets?

05-Jan-2017 Intellasia | Forbes | 6:00 AM Print This Post

Have you ever searched online for ‘house price’ in China? Our anecdotal evidence suggests that in some cities at some points in the cycle analysing the volumes of searches conducted online might be a useful predictive indicator of price spikes.

Before slowing down in October and November after the introduction of new home purchase restrictions in the early October, the Chinese house prices in several major cities had experienced a significant run-up, as the overall index shows on the chart below, capturing the sequential month-on-month price changes.

For individual cities, the acceleration of price growth happened at different points in time. Here are two examples looking at China’s top tier cities Shenzhen and Nanjing.

What could have helped predict such spikes in prices?

While we were mining the data for some other purpose at my company, we accidentally encountered an interesting relationship between the price growth (here based on data from SouFun-CREIS) and the index of online searches for ‘??’ or ‘house price’ in some cities, based on Baidu Index (noting that Google does not operate in China). Such a search volume index measures a relative number of searches conducted on a particular phrase and when viewed over time, can indicate relative spikes and drops in interest in a particular topic.

Cross-referencing the search volume indices above and the house price growth data shows that in at least these cases there were clear spikes in the search volume indices just before major run-ups in prices. Notice how the search volume index charts above show a spike in the index at the end of May 2015 and mid-January 2016 for Shenzhen and in mid-January and early March 2016 for Nanjing, preceding the price spikes in March-April 2016 for Nanjing and June-July 2015 and January-February 2016 for Shenzhen.

There is a common sense explanation why it would work. As markets heat up, more people start considering jumping in, start making online checks, and when that turns into a rush to buy (rationality aside), then prices escalate until the proverbial music stops.

Many academic papers have been written on the topic of predictive powers of search volume indices (you can google the term), especially on the US data. In China’s property markets, I’m aware of one that tried to more generally predict price movements through search volume indices in Beijing and Xi’an: ‘A study on a correlation between web search data and housing price. Evidence from Beijing and Xi’an in China’.

The conclusion there was that ‘(…) the predictability of web search data has a certain relationship with economic development level of the region, and web search data has a better explanation ability for the fluctuation of housing price in a developed region, like Beijing.’

In another paper, ‘Forecasting Chinese Tourist Volumes With Search Engine Data’, authors look at predictive powers of online searches for visitor volumes in Hainan province in China.

It concluded that ‘(…) compared to traditional methods of monitoring visitor numbers, the predictive power of the search data models in this study was much higher’.

For our cases on China house prices, looking at a few cities, it’s only anecdotal with a short lead time. Yet it shows some alternative indicators may well be worth a closer watch.


Category: China

Print This Post

Comments are closed.