Research

Work in Progress

Experiments in Information Acquisition and Voting
(with S. Bandyopadhyay, J. Lohse, R. McDonald)

Is majority voting an effective way of aggregating information when voters must acquire information from sources of varying quality? We study this question in a lab experiment, where participants can obtain information by choosing between two sources, which are imperfectly informative. The less informative source grants a ‘consumption benefit’. We compare four information environments between subjects. We find that in “source and signal” treatment, in which the signal source choice and the signal realisation of group-members are both observable, participants acquire low quality information significantly more often. This does not harm their outcomes since this treatment still yields a high average payoff per person. In sum, we present the first evidence regarding the effect of acting in various information environments, on endogenous information acquisition, voting and payoffs.

Information Acquisition in a Threshold Public Goods Game
(with S. Bandyopadhyay, J. Lohse, R. McDonald)

Threshold uncertainty has been previously shown to impede the efficient provision of threshold public goods. Do groups perform better when their members can acquire information on the location of the threshold? We conduct an online experiment on threshold public good provision where participants can simultaneously choose to buy private clues about the location of the threshold. Each group contains two low-risk and two high-risk members who differ in their expected loss from failing to reach the threshold. We find that groups reach the threshold less frequently when group members can acquire information than when no information acquisition is possible. We attribute this to optimism: individuals acquire too little information to adequately update their priors about the true threshold value, resulting in too low contributions. Low-risk members are disproportionately responsible for the failure of collective action. Our findings suggest that giving individuals a possibility to learn via private information acquisition is insufficient in solving the problem of threshold uncertainty.

The Impact of Social Media on Traditional Media News Consumption
(with S. Vaidya)

Browsing information through social media via posts by friends and family on platforms like Facebook and Twitter has become an integral part of modern living. This access is naturally impacting on people’s news consumption habits with flow-on impact on traditional news media companies who are increasingly having to put up pay-walls amongst diminishing advertising revenues. In this paper, we analyse this issue via a theoretical model by examining a potential consumer’s decision on whether to pay to consume traditional news in the presence of free information available on social media. We find that a consumer’s prior (or bias) about a state of interest, the access-cost of traditional media, as well as the relative inaccuracy of information on social media compared to traditional media news all impact the consumer’s decision to use traditional media. Surprisingly, free information on social media need not always act as a substitute to news consumption on traditional media. Indeed we find that when social media accuracy and cost of access to traditional media are relatively low, there could be circumstances where social media increases the consumption of traditional media by enhancing the value of traditional media information for some readers. Indeed this effect can persist even if social media becomes incrementally more informative via greater fact-checking. However if greater social media accuracy is coupled with higher costs of accessing traditional media, then the reader composition of traditional media might change towards more opinionated readers. Once the access cost of traditional media is sufficiently high, social media acts as a substitute and crowds out the demand for traditional media.

Effective Altruism and Peer-to-peer Payment Apps
(with H. Mohtashami Borzadaran)

Can peer-to-peer (P2P) mobile payment applications be a valuable tool in the drive for charitable fundraising? We propose to investigate this question in an online experiment, where we carefully select groups of ‘friends’ on a social media platform and provide them with suitable monetary incentives, to mimic a P2P payment environment. P2P apps present a new avenue for fundraising by eliciting prompt repayments through them with the guarantee of donating a portion of that money to charity. There are several factors combined in these apps that could work favourably for fundraising: 1) the nudge for repayment of money owed – a unique feature not found in other fundraising methods, 2) digital transactions – crucial in the post-pandemic world, 3) verified and pre-specified charitable organisations – boosts trust and serves the goal of effective altruism, 4) caring about the cause that is meaningful to the member of one’s social group who is soliciting these donations – triggers relational warm glow, 5) the fact that one’s contribution is visible to the other person – identity-signalling and reputation concerns are at play here, and 6) ease of donating – day-to-day transactions that make these apps popular, are coupled with accessible charitable-giving. Thus, given all these advantages, it is imperative to understand whether we can successfully harness the potential power of these apps for charitable fundraising. To this end, we run two treatments and compare them against our baseline to find the most effective P2P payment framework. In the baseline (Base), we create a simple framework of P2P apps – repayment to friends. Our first treatment (Altruism) includes automatic donation of a fraction of the repayment to a charity of choice, which would be common knowledge to both parties. In our second treatment (Social Image), we study the impact of social image on repayments, by adding a sharing option. Here, donations can be immediately publicised to one’s wider social media community. This treatment would tell us whether the additional factor of recognition proves to be an important extrinsic motivator. We predict that as compared to Base (Altruism), Altruism (Social Image) would increase the number of repayments and decrease the repayment times, thereby resulting in more donations.