Deriving economic insights from Twitter and Spotify

Social media is having an emerging impact on markets and trading that is evolving faster than analysts can identify. And Donald Trump’s recent tweets are no exception.

At the start of May, Trump tweeted suggestions that trade negotiations with China were progressing too slowly, sparking fear of a trade war for investors.[1] This was followed up by a tweet that both “misstated” and “mischaracterised” US-China trade relations.[2] However, Trump’s claims were enough to send the markets tumbling with the Dow plummeting by 450 points on Monday. This emphasises the emerging impact social media is having on investment.

This pattern also reflects a wider shift at the nexus of economics and emotions. Just as Trump’s tweets, which include prescriptive emotional connotations, are influencing investment behaviour, deciphering emotional patterns across social media provide insights into wider economic activity. As such, social media is being used to capture consumer sentiment and spending habits to more effectively conceptualise public policy and marketing campaigns.

Emoji tracking allows brands to target advertising campaigns to specific users by determining those who are more likely to spend money on their products. Using simplistic principles, emojis are narrowing advertisement focus. Brand names which are tweeted alongside ‘love heart eye’ emojis reveal emotional connections users have with products, signalling a greater potential for targeted advertising campaigns.[3] Furthermore, marketing companies have been targeting fast food adverts to those who tweet the pizza emoji, whilst displaying athletic wear advertisements to users who tweet footballs.

Emojis are also providing more information to advertisers beyond the variables of age or gender which are linked to Twitter accounts. The use of emojis with different skin tones is providing advertisers with the ability to target adverts to people of different races.[4]

Targeted advertising is estimated to be three-times more effective than mass campaigns, generating greater interaction, likes and retweets.[5] This extends the text-understanding capabilities of algorithms beyond Twitter’s 280-character limit, making emojis indicators worth a thousand words. Furthermore, some companies are creating variations of their advertisements, with Toyota releasing 83 versions of one ad which is presented to users based on their recent emoji usage.[6] Yet, it is not only the private sphere that is benefitting from analysing online activity.

The Bank of England is also leveraging social media to gain an insight into national economic sentiment. Andy Haldane, the BOE Chief Economist, has emphasised the utility of capturing music trends on Spotify to infer consumer confidence levels and inform interest rate changes.[7] This could provide a more robust understanding of the implications of monetary policy actions.

The emotional valence of music can be calculated by analysing the emotional loading of a song against a series of eight emotions. Joy, sadness, anger, fear, disgust, surprise, trust and anticipation are all core components of an algorithm that promises to decipher Spotify Charts. Research from Claremont Graduate University suggests that algorithms used to measure the emotional valence of Spotify derive the same results as conventional consumer confidence surveys.[8] This has major implications on how information about consumer confidence is gathered due to progressive improvements in the predictive abilities of algorithms.

Similarly, this technique has been used in the past to predict investment patterns using Twitter, with one study predicting the Dow Jones Industrial Average with 86.7% accuracy.[9] Through measuring the collective mood states derived from large-scale Twitter feeds, researchers had enough information to reliably predict investment activity.

Ethical implications

In spite of their power, these techniques are garnering criticism from users who refer to this process as psychological profiling which manipulates personal information. Furthermore, deriving information such as race, which is not connected to online profiles, is raising questions around the ethics of analysing online information. Using information encoded in tweets and music popularity might be the first step towards an Orwellian future, yet big data agencies suggest that this practice is in accordance with the terms of service of online platforms.[10]

Social media is allowing emotions to inform economic decision making and heralds a new age in the modern economy. Emotional tweets underscore the movements of investors and reveal patterns for efficient marketing practices, whilst Spotify can create a snapshot of aggregate economic sentiment. This emphasis on emotions means that predictable behaviour aligned with the textbook homo economicus may become less prevalent as we become increasingly irrational, which is why algorithms may be able to better capture these patterns.

The online sphere is evolving so rapidly that it is hard to isolate its impacts. This represents a shift in economic information which sits alongside other vital economic indicators as signifiers for investment and economic activity. As the digital sphere continues to expand faster than legislation can keep up, it seems that tracking Twitter and Spotify habits will continue into the future.

[1] https://twitter.com/realDonaldTrump/status/1125356705787850753

[2] Yglesias, Matthew. (2019). “The Trump trade tweets that sent the stock market tumbling, explained”. Retrieved from https://www.vox.com/2019/5/6/18531101/trump-china-tariff-tweet-schumer-stock

[3] The Times. (2018). “Why our Twitter emojis give advertisers big smiling faces”. Retrieved from http://link.galegroup.com/apps/doc/A566128343/STND?u=monash&sid=STND&xid=49c4731f.   

[4] Balonon-Rosen, Peter. (2018). “That emoji you just tweeted could determine the next ad you see”. Retrieved from https://www.marketplace.org/2018/02/16/world/emojis-change-how-we-communicate-advertisers-track

[5] The Times. (2018). “Why our Twitter emojis give advertisers big smiling faces”. Retrieved from http://link.galegroup.com/apps/doc/A566128343/STND?u=monash&sid=STND&xid=49c4731f.  

[6] Balonon-Rosen, Peter. (2018). “That emoji you just tweeted could determine the next ad you see”. Retrieved from https://www.marketplace.org/2018/02/16/world/emojis-change-how-we-communicate-advertisers-track

[7] Kaivanto, Kim and Zhang, Peng. (2018). “Your Spotify history could help predict what’s going on in the economy”. Retrieved from https://theconversation.com/your-spotify-history-could-help-predict-whats-going-on-with-the-economy-96305

[8] Sabouni, Hisam. (2018). “The rhythm of markets”. Retrieved from https://www.researchgate.net/publication/323560860_The_Rhythm_of_Markets

[9] Bollen, J., Mao, H., & Zeng, X. (2011). Twitter mood predicts the stock market. Journal of computational science, 2(1), 1-8.

[10] Balonon-Rosen, Peter. (2018). “That emoji you just tweeted could determine the next ad you see”. Retrieved from https://www.marketplace.org/2018/02/16/world/emojis-change-how-we-communicate-advertisers-track