Google Econometrics

Anonymous1987

The following may interest some

The current economic crisis requires fast information to predict economic behavior early, which is difficult at times of structural changes. This paper suggests an innovative new method of using data on internet activity for that purpose. It demonstrates strong correlations between keyword searches and unemployment rates using monthly German data and exhibits a strong potential for the method used.

http://www.atypon-link.com/DH/doi/pdf/10.3790/aeq.55.2.107

Essentially Google searches are shown to be accurate at predicting unemployment, a fascinating development with numerous applications for those data starved economists. What are everyone's thoughts on this?

andrew

The dataset they used can be accessed here. I wonder could this be used to predict recessions too; I presume finding keywords pertaining to GDP growth would be really hard. If it was possible to access this data in real time, could unemployment rates be ascertained in pretty much real time?

Anonymous1987

andrew wrote: »

The dataset they used can be accessed here. I wonder could this be used to predict recessions too; I presume finding keywords pertaining to GDP growth would be really hard. If it was possible to access this data in real time, could unemployment rates be ascertained in pretty much real time?

Effectively yes though we already have a real time predictor of recessions with financial market data, specifically the yield curve. However, there are numerous other applications which could be very interesting. See this paper for a description of the Google price index giving real time information which normally takes months to produce. The information is not without its limitations, clearly using data relating exclusively to internet searches is one and it is important that when interpreting the data that only predictions about the present can be made. This is still an improvement with regard to the delays in receiving market information.

later12

I wonder when markets will start using this, and googles domestic trends as a real time indicator of economic data to pre-empt the outcome of official reports, fascinating stuff. A similar analysis of twitter and perhaps facebook trends could prove similarly intriguing.

macannrb

later10 wrote: »

I wonder when markets will start using this, and googles domestic trends as a real time indicator of economic data to pre-empt the outcome of official reports, fascinating stuff. A similar analysis of twitter and perhaps facebook trends could prove similarly intriguing.

I think they have already started using similar methods in bots, so I assume this is already being used to some degree



http://ftalphaville.ft.com/blog/2011/03/28/528481/for-the-bots-anne-hathaway-is-not-warren-buffett/

meryem

I see a future wherein rogue bots will try to suck the bandwidth of even ordinary sites by uselessly indexing their not for indexing pages even. Yeah in the search of financial data to make economic gain. Google can give them some directions by guiding which kind of sites should be targeted by it's trends site.

andrew

Surely if a site can't handle Google indexing it's pages, it's got bigger problems than the fact that Google is indexing it's pages?

later12

Read an interesting piece on this is the FT this morning (pg 4 hard copy)



http://www.ft.com/intl/cms/s/0/f9550984-952e-11e0-a648-00144feab49a.html#axzz1PGbSNii7

The data could be helpful in producing the Bank’s so-called “nowcast” – a prediction of the current state of the economy, based on the latest available information. The Bank uses a “nowcast” because official data from the Office of National Statistics are often slow to appear and the Bank does not always trust some early estimates of data because they are frequently revised...

Derwent Capital Markets, a family-owned investment boutique, said last month that it had launched a hedge fund using Twitter and other social network postings to predict the financial markets.

The ONS has produced an analysis of the predictive power of search terms for retail sales, and found it to be a good match for some elements of shoppers’ spending. Central banks in Italy and Israel have also looked at internet searches as a possible guide to economic activity.

edit: that was actually yesterday's FT. They all look the same.

andrew

Speaking (tangentially) of nowcasting, here's a paper which suggests that nowcasting is significantly superior to forcasts produced by human forcasters. Which probably isn't that surprising really.