Twitter is a tried and tested technique for time-pressed journalists at less serious publications than GlobalCapital to source a quote for a news story, or even to get inspiration for a new article.
Now regulators may be about to discover the joys and pitfalls of the platform.
Using Twitter data gives a more accurate forecast of depositor discipline than a standard model, according to researchers at the Bank of Italy. Using the data is particularly effective for weaker banks, they found.
The findings were published in February by Matteo Accornero and Mirko Moscatelli in a paper entitled “Listening to the buzz: social media sentiment and retail depositors’ trust”.
They trawled through tweets, analysing what the words said about the sentiment of the bank in question.
Twitter as a source of information provides a distinct benefit compared with other types of more conventional financial data, they reckoned: “Rumours spreading over the internet possess two important advantages over financial data: they are available on a continuous basis and they express the sentiment of a population of users normally not necessarily interested in the financial sector.”
It could play a role in the work of bank supervisors.
“Twitter has a potential application in helping to assess, rather than the health of a bank (for which we have better resources), the ‘trust’ that people have in that bank,” the authors told GlobalCapital.
And it is not just in an analysis of retail deposits where Twitter could be handy. Other researchers see a link between tweets and a bank’s financial ratios.
Giuseppe Bruno and Juri Marcucci from the Bank of Italy along with Paola Cerchiello and Giancarlo Nicola from the University of Piva used tweets from Italy to measure investor sentiment about Banca Monte dei Paschi, UniCredit, Intesa Sanpaolo and Deutsche Bank.
They found evidence of causality in both directions: financial indicators affected Twitter sentiment, and vice versa. Correlation was stronger for the banks which incite the most "buzz" — Monte dei Paschi and Deutsche Bank.
“Twitter sentiment positively affects the stock returns and volume of traded stocks, while it seems to negatively affect the CDS spreads,” the researchers said in a presentation.
But difficulties encountered in the study included the shortness of tweets and finding the tweets themselves (searching for Monte dei Paschi using its acronym brought up tweets about UK members of parliament).
‘Fake news’
Fake Twitter rumours having real-world consequences on the market are nothing new either.
In August 2012, a Twitter account claiming to belong to Vladimir Kolokoltsev, then the Russian interior minister, tweeted that Bashar al-Assad had been killed or injured, then confirmed he had died.
The first tweet was posted at 9:59am, and between 10:15am and 10:45am futures of light, sweet crude rose from $90.82 to $91.99 a barrel on the New York Mercantile Exchange, according to the Wall Street Journal.
Meanwhile, Twitter users have already tried to attack banks. When CaixaBank and Sabadell decided to move their headquarters out of Catalonia in October last year after the region’s independence referendum, Catalan activists called for depositors to withdraw cash in a coordinated fashion.
One of those tweets came from the region’s National Assembly, which has 357,000 followers at the time of writing.
How much this mattered is unclear: Sabadell went from €100.5bn in deposits to €98.7bn over October, but CaixaBank actually saw a rise in deposits from €192.2bn to €192.7bn.
But there remains a chance that activists can easily gain traction with their messages on Twitter, causing damage to banks and misleading regulators at the same time.
The authors of the paper on retail depositors played this down.
“Soft data will unlikely (at least, surely not in the near future) substitute hard financial data, and it will not generate by itself a strong distress signal,” they told GlobalCapital.
“The idea is that, as social media data is more timely and heterogeneous, it can give some early-warning alert that can then be verified using more reliable data,” they said
But others see a risk. Deutsche Bank has decided to focus on proper financial news rather than social media data in its recently launched an artificial intelligence tool to sift through text.
“Using financial media is, of course, far better than social media, far more viable as a dataset for this kind of analysis,” said Spyros Mesomeris, global head of quantitative strategy at the bank.
“With social media, you have to be very careful about who is tweeting and about retweeting. Even if the information is useful at first, the information decay can be quite sharp,” he said.
A journalist would of course argue that financial media is a more reliable source of information than the chatter of the masses on Twitter.
But market observers should take care: the financial world is not immune to fake news either.