Social media is an increasingly important channel for communication and information dissemination. But social media is also a channel for firms to persuade customers to buy their products. In the context of financial markets, firms can use social media to engage with current and potential investors. While other means of communication between firms and investors exist (mandatory and voluntary disclosures), communication via social media is less constrained by regulations, takes place in real time, and is bidirectional.
In BSE Working Paper 1366, “Tweeting for Money: Social Media and Mutual Fund Flows,” Javier Gil-Bazo and Juan F. Imbet explore persuasion in social media by asset management firms. They focus on the mutual fund families that manage domestic equity funds in the U.S. and gather more than 1.5 million tweets from 284 mutual fund management companies’ accounts to investigate whether the firms use social media to attract investors’ money.
Associative thinking, persuasion, and who tends to tweet more?
Persuasion is related to how individuals evaluate each message. Humans tend to have an associative way of thinking. Namely, we evaluate several objects by grouping them into categories. Firms can use associative thinking and social media to persuade their customers. One example is in financial markets. During high investor sentiment times, asset management firms can tweet about enhancing opportunities. When investor sentiment is low, they can use social media to position themselves as trustworthy advisers. Due to associative thinking, investors may relate such preferred characteristics to the product even if communications lack real informational value.
Not every asset manager will succeed in persuasion using social media. Unlike advertising, social media is not one-way communication. Users can react to the messages of the mutual fund family and criticize them. And these reactions will be observed and amplified multiple times by other users.
Managing social media well requires coordination between the social media manager and the marketing department. Plans and goals must be made to engage customers. The larger the firms are, the more capacity they will have to use social media for persuasion. Larger firms with more resources are not the only firms that want to engage customers through social media. Younger firms may have more tendency to use social media to gain recognition. The authors start by investigating which factors affect the mutual funds’ decision to use Twitter. They find that economies of scale play a key role. Larger mutual fund families tend to have Twitter accounts. Among the families with the accounts, larger, younger families with better performance and lower risk tweet more than others.
Tweets with a positive tone can increase more flow of money to the mutual funds
Whether tweets from mutual fund families have some impact on investors is another question that the authors have investigated. Tweets can attract flows of money from investors if tweeting has a persuasive impact on investors. The authors use supervised machine learning algorithms on the almost 1.6 million gathered tweets to classify the tones as positive, negative, or neutral. Figure 1 shows the number of tweets from mutual fund families. The authors find that more positive tweets by a mutual fund family are followed by higher inflows of money from investors to the family’s funds. To be precise, an increase of one standard deviation in the positiveness of the tone of tweets during one month is associated with an average increase in assets under management of nine basis points in the following month, which corresponds to 7.25 USD million for the average family.
Tweets regarding financial advice and market commentary have more impact.
There can be many explanations regarding the mechanism of such impact. On the one hand, mutual fund families can implicitly associate their products by tweeting about financial advice or commenting about the current market situation. On the other hand, social media can be another channel for firms to provide customer service. To explain the relationship between positiveness and inflows, the authors use Latent Dirichlet Allocation (LDA). With this topic modeling algorithm, the tweets are classified into three topics which are 1. Customer service, 2. Market commentary, and 3. Financial advice. Figure 2 shows the word clouds for these three categories with the size representing how often each word is used.
The results suggest that positiveness of tweets in all category leads to an increase in money inflows. However, positiveness in market commentary and financial advice has a more significant impact compared to customer service. Hence, the inflows from investors cannot be explained just by their appreciation for the service. The authors also find that positiveness’s impact is greater during the high-sentiment time. This is because they focus on equity mutual funds, which are riskier than bond mutual funds. Persuasion during high time leads customers to seek investment products with higher risks and higher returns.
A high-frequency approach
Although in their tests, the authors control for a variety of observable fund flow determinants as well as unobservable time-invariant fund and fund family characteristics, they cannot discard that tweets respond to unobservable news that, in turn, predict fund flows. To identify the effect of asset management companies’ tweets on investor decisions, they combine a high-frequency approach with intraday ETF trading data. More specifically, the authors study whether an ETF’s price changes in a very short time window around the time when the ETF’s family tweets. This high-frequency approach allows for clean identification of the effect of tweets on prices with the identifying assumption being that over such a short window of time no other relevant information affecting the security’s price is released. The results indicate that a family’s tweets with a positive tone can push up the prices of its ETFs’ shares. Consistently with the effect being amplified by retweeting and reacting, the effect on ETF share prices doubles in the range of 10 minutes after posting the tweets. However, since tweets contain no fundamental information, the effect subsides within 35 minutes.
Finally, the authors test the alternative explanation that asset management companies use social media to alleviate information asymmetries by either reducing search costs or conveying privately observed information. The tests reject this hypothesis. Positive tweets not only increase inflows of funds, but also reduce outflows, which is inconsistent with the search cost hypothesis. Also, positiveness does not appear to contain information about superior future fund performance.
Social media, such as Twitter is a great tool for asset management firms to communicate and provide their customers with easier-to-access information. The results from this study suggest that mutual funds families have more incentives to use social media as a tool for persuasion rather than to inform investors.