Vladislav Dombrovskyy, Ph.D., Associate Professor,
Ukrainian Academy of Banking of the National Bank of Ukraine, Ukraine
USING GOOGLE TRENDS FOR BETTER MONETARY POLICY IN EMERGING ECONOMIES
Timely economic statistics is crucial for effective decision making, but it is released with a lag by national statistical services. Finding data that contains information regarding the current economic situation is especially valuable for emerging economies, which generally experience bigger delays in term of official statistics availability. We looked at using Google search query data for nowcasting unemployment rate in Ukraine and found strong correlation between keyword searches related to labor market and unemployment rates. Google data is especially useful for predicting the turning points of a time series.
There are several benefits from using search queries data for improving economic forecasts: 1) They are derived directly from households; 2) It is a far more representative sample than results of consumer surveys; 3) They released at high frequency and regular intervals.
However, there are two main prerequisites for using search data as a source of information: 1) High level of the Internet penetration is a must; 2) Google is used as a default search engine by most users in a country, because there is no other search engine which provides access to search statistics. Only under these conditions, the search data is reliable.
Ukraine is ready for adopting this approach, as it has about 20 million active internet users in 2012. The penetration level was about 45 %, which is relatively lower comparing to developed countries (e.g. the U.S. 78 %), but the research showed that this level is sufficient.
Nevertheless, using data from the Internet for making economic decisions creates a few potential problems. First of all, it would be possible to influence economic policy of any given country by using Internet bots for producing misleading search queries. Secondly, with enough data and computational power it is easy to find something completely irrelevant that plausibly correlates with main macroeconomic indicators. Thirdly, we have to take into account reflexivity in economics. If people knew that their search queries influence governments’ decisions, they could try to manipulate them.