Twitter is a goldmine for tracking consumers’ moods on prices, according to findings of the Bank of Italy | Instant News


ROME (Reuters) – The Bank of Italy on Monday said its experimental set of indicators from the content of millions of tweets accurately tracks consumers’ moods on prices, offering scope for a new powerful monetary policy tool.

The effort comes as economists and policymakers around the world increasingly turn to social media and other unconventional sources to measure consumer behavior and as inflation continues to exceed targets set by many of the leading central banks.

Researchers found their indicator, based on millions of tweets, was calculated not only by the final inflation readings and existing price expectation measures by Italy’s national statistical offices, financial markets and other forecasters, but also in real time and provided more detailed detail.

“The results suggest that Twitter can be a new, timely resource for devising methods of gaining trust,” said the authors of the 107-page study, adding they believe Italy-focused research can be replicated elsewhere.

Twitter has about 200 million monthly active users worldwide and had about 10 million active users in Italy as of 2019, the authors said.

The analysis began by collecting 11.1 million tweets posted in Italian between June 2013 and December 2019 that contained at least one of a pre-selected set of words related to inflation, prices and price dynamics.

“The reason for focusing on the purely raw tweet count is the intuitive idea that the more people talk about something, the more likely it is it reflects their opinion and that their views can influence other people’s expectations,” he said.

Then the dataset is “cleaned” to remove advertisements or tweets that use the word inflation in an unrelated context.

In this way, a tweet like “#Draghi: ‘We are saving Europe from deflation.’ Don’t count your chickens before they hatch! “Are kept, while others, such as” Only in Baby Glamor if you buy the three cheapest items for free. Promotional sales until October 10 ”are filtered.

The remaining dataset is used to construct two indexes based on expected increase or decrease in inflation by measuring the daily volume of tweets containing pre-selected word combinations such as “low price” or “very high price”.

“The fact that economic agents are talking about expensive bills should reflect higher inflation expectations,” the report said. “On the other hand, the people discussing falling oil prices should match expectations of lower inflation.”

The final set of indicators is then created based on the divergence between the two indices.

The authors say their work underscores the significance and policy implications of the information contained on social networks but acknowledges that further study is needed to interpret the data.

They also note that there have been cases of Twitter-based indicators being thrown off by viral social media events, for example when a record $ 236 million apartment sale in 2014 led to multiple tweets containing variants of the phrase “more expensive”.

Edited by Mark John and Raissa Kasolowsky

.



image source

to request modification Contact us at Here or [email protected]