Economic Value of Data: Quantification Using Online Experiments

Abstract

We conduct incentive compatible choice experiments to measure the economic value of social media data. We focus on the value that users place on their personal data related to the three biggest social media platforms: Facebook, Instagram and Twitter. We find that the median Willingness to Accept (WTA) of users for the entire "stock" of data for Facebook and Instagram is $300 each which is 20% higher than Twitter valued at $250. We use data valuations from a recent data breach settlement by Facebook to provide information interventions. We find that reducing these information frictions about the value of data can make users revise their valuations upwards. They do not

Choice Architecture, Privacy Valuations, and Selection Bias in Consumer Data

“…On the other hand, the effect of price anchor is negatively correlated with the average valuation: consumers who have lower privacy valuations are also more likely swayed by a price anchor (correlation is -0.46). This pattern is consistent with Collis et al (2020), who find that consumers who receive an informative signal on the data value often revise their valuation upwards when their prior is lower than the signal, but rarely update their belief downwards when their prior is high.…”

“…Focusing on log valuations makes the results robust to the impacts of extremely high values and our top-code choices. We note that the long-tail pattern in privacy valuations is universal (e.g., see Collis et al 2020 andLin 2022) and is not unique to our setting.…”

“…These heterogeneity results are broadly in line with findings in the existing literature. Regarding privacy valuations, Goldfarb & Tucker (2012) find that older people and women value their privacy more, and Collis et al (2020) show that high-income consumers and Asian communities value their Facebook data more. As to frame effects, Mrkva et al (2021) similarly find that lower socioeconomic status predicts stronger choice architecture effects.…”

“…Although existing literature has documented the presence of privacy preferences, empirical studies quantifying the value of data to consumers are nascent (Acquisti et al 2013, Lin 2022, Tang 2019, Collis et al 2020. One potential reason is that privacy preferences are context-specific (Martin & Nissenbaum 2016) and hard to measure.…”