Abstract
We consider a platform providing free content for users and earning profit from the sale of advertising. The platform can collect and analyze personal data to customize advertisements for individual users. Customization may alleviate users’ aversion to advertising, but it may also raise privacy concerns. Considering the platform as a two-sided market with asymmetric externalities, we investigate the effects of privacy concerns on the platform’s optimal advertising pricing and data collection strategies. We find that it is always beneficial for the platform to collect and analyze personal data. When users attach less concerns on privacy, the price increases with the efficiency of efforts and decreases with the initial nuisance cost of advertisements to users. However, if users attach more concerns on privacy, the price decreases with the efficiency of efforts and increases with the initial nuisance cost of advertisements to users.
Similar content being viewed by others
References
Lv, D., & Zhu, S. (2019). Achieving correlated differential privacy of big data publication. Computers & Security, 82, 184–195.
Ghasemaghaei, M., & Calic, G. (2019). Can big data improve firm decision quality? The role of data quality and data diagnosticity. Decision Support Systems, 120, 38–49.
Acquisti, A., Taylor, C., & Wagman, L. (2016). The economics of privacy. Journal of Economic Literature, 54(2), 442–492.
Montes, R., Sand-Zantman, W., & Valletti, T. (2018). The value of personal information in online markets with endogenous privacy. Management Science, 65(3), 1342–1362.
Valletti, T., & Wu, J. (2020). User profiling with data requirements: Structure and policy implications. Production and Operations Management, 29(2), 309–329.
Cohen, M. C. (2017). Big data and service operations. Production and Operations Management, 27(9), 1709–1723.
Kox, H., Straathof, B., & Zwart, G. (2017). Targeted advertising, platform competition, and privacy. Journal of Economics & Management Strategy, 26(3), 557–570.
Ghose, A., & Todri, V. (2016). Towards a digital attribution model: Measuring the impact of display advertising on online user behavior. MIS Quarterly, 40(4), 889–910.
Goldfarb, A., & Tucker, C. (2011). Online display advertising: Targeting and obtrusiveness. Marketing Science, 30(3), 389–404.
Hao, L., Guo, H., & Easley, R. (2016). A mobile platform’s in-app advertising contract under agency pricing for app sales. Production and Operations Management, 26(2), 189–202.
Liu, J., Liu, F., & Ansari, N. (2014). Monitoring and analyzing big traffic data of a large-scale cellular network with Hadoop. IEEE Network, 28(4), 32–39.
Bi, S., Zhang, R., Ding, Z., et al. (2015). Wireless communications in the era of big data. IEEE communications magazine, 53(10), 190–199.
He, Y., Yu, F. R., Zhao, N., et al. (2016). Big data analytics in mobile cellular networks. IEEE access, 4, 1985–1996.
Ghemawat, S., Gobioff, H., & Leung, S. T. (2003). The Google file system. Proceedings of the nineteenth ACM symposium on Operating systems principles, pp. 29–43.
Dean, J., & Ghemawat, S. (2008). MapReduce: Simplified data processing on large clusters. Communications of the ACM, 51(1), 107–113.
Hadoop, A. (2011). Apache hadoop. https://hadoop.apache.org.
Li, C., Li, D. Y., Miklau, G., et al. (2014). A theory of pricing private data. ACM Transactions on Database Systems (TODS), 39(4), 34.
Shen, Y., Guo, B., Shen, Y., et al. (2016). A pricing model for big personal data. Tsinghua Science and Technology, 21(5), 482–490.
Ruiming, T., Huayu, W., Zhifeng, B., et al. (2013). The price is right: Models and algorithms for pricing data. In DEXA'2013: 24th international conference on database and expert systems applications (pp. 380–394) Springer.
Spiekermann, S., Acquisti, A., Böhme, R., et al. (2015). The challenges of personal data markets and privacy. Electronic markets, 25(2), 161–167.
Purcell, K., Rainie, L., & Brenner, J. (2012). Search engine use 2012. https://www.pewresearch.org/internet/2012/03/09/search-engine-use-2012/.
Kummer, M., & Schulte, P. (2019). When private information settles the bill: Money and privacy in Google’s market for smartphone applications. Management Science, 65(8), 3470–3494.
Wu, K. W., Huang, S. Y., Yen, D. C., et al. (2012). The effect of online privacy policy on user privacy concern and trust. Computers in Human Behavior, 28(3), 889–897.
Bennett, S. C. (2010). Regulating online behavioral advertising. J. Marshall L. Rev., 44, 899.
Tang, Z., Hu, Y., & Smith, M. D. (2008). Gaining trust through online privacy protection: Self-regulation, mandatory standards, or caveat emptor. Journal of Management Information Systems, 24(4), 153–173.
Tucker, C. E. (2014). Social networks, personalized advertising, and privacy controls. Journal of Marketing Research, 51(5), 546–562.
Lee, Y., & Kwon, O. (2010). An index-based privacy preserving service trigger in context-aware computing environments. Expert systems with applications, 37(7), 5192–5200.
Casadesus-Masanell, R., & Hervas-Drane, A. (2015). Competing with privacy. Management Science, 61(1), 229–246.
Rochet, J. C., & Tirole, J. (2006). Two-sided markets: A progress report. The RAND Journal of Economics, 37(3), 645–667.
Armstrong, M. (2006). Competition in two-sided markets. The RAND Journal of Economics, 37(3), 668–691.
Kaiser, U., & Wright, J. (2006). Price structure in two-sided markets: Evidence from the magazine industry. International Journal of Industrial Organization, 24(1), 1–28.
Ferrando, J., Gabszewicz, J. J., Laussel, D., et al. (2008). Intermarket network externalities and competition: An application to the media industry. International Journal of Economic Theory, 4(3), 357–379.
Roger, G. (2017). Two-sided competition with vertical differentiation. Journal of Economics, 120(3), 193–217.
Kumar, S., & Sethi, S. P. (2009). Dynamic pricing and advertising for web content providers. European Journal of Operational Research, 197(3), 924–944.
Carroni, E. (2018). Behaviour-based price discrimination with cross-group externalities. Journal of Economics, 125(2), 137–157.
Reisinger, M. (2012). Platform competition for advertisers and users in media markets. International Journal of Industrial Organization, 30(2), 243–252.
Cho, C. H. (2004). Why do people avoid advertising on the internet? Journal of Advertising, 33(4), 89–97.
Gal-Or, E., Gal-Or, R., & Penmetsa, N. (2018). The role of user privacy concerns in shaping competition among platforms. Information Systems Research, 29(3), 698–722.
Acknowledgements
The authors gratefully acknowledge the support from the National Natural Science Foundation of China (71771179, 71532015) and the International Exchange Program for Graduate Students of Tongji University.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interests
The authors declare that there is no conflict of interests regarding the publication of this article.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
Proof of Proposition 1
Differentiating Function (8) with respect to \(p,e\), we obtain:
By the first order condition \(\frac{{\partial \pi_{p} }}{\partial p} = 0,\frac{{\partial \pi_{p} }}{\partial e} = 0\), we have
and \(e^{*}\) satisfies
In the quadratic Eq. (12),
And if \(k \ge \frac{{2\alpha^{3} \beta^{2} }}{{(1 + \alpha \delta )^{2} }} - 2\alpha \gamma\), it follows that
Since \(k + 2\alpha \gamma > 0\), we have \(\left( {k + 2\alpha \gamma } \right)^{2} (1 + \alpha \gamma )^{2} - 2\alpha^{3} \beta^{2} \left( {k + 2\alpha \gamma } \right) \ge 0\). As a result, \(\Delta > 0\), there are two different solutions of Eq. (12).
By solving the quadratic Eq. (12), the two solutions to (12) are as following: \(e_{1} = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - 2\sqrt \Delta }}{{3\alpha \beta \left( {k + 2\alpha \gamma } \right)}}\), \(e_{2} = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) + 2\sqrt \Delta }}{{3\alpha \beta \left( {k + 2\alpha \gamma } \right)}}\).
Next, we will prove that \(e_{2}\) is not a feasible solution to the problem.
If \(e = e_{2}\), then
Since
And from Eq. (14), it follows that \(\frac{2}{{9\alpha \beta^{2} }}\left( {2\alpha^{3} \beta^{2} - \left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} } \right) \le 0\). Hence, Eq. (16) is non-positive, \(\left( {\frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}} \right)^{2} - \left( {\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }}} \right) \le 0\). Thus, \(\left| {\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }}} \right| \ge \left| {\frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}} \right|\). Because \(\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }} > 0\), \(\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }} \ge \left| {\frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}} \right| \ge \frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}\). Hence, from Eq. (15), we have \(p - \frac{{ke^{2} }}{2} \le 0\). This implies that the profit of the platform from each advertisement is nonpositive. Obviously, it is unreasonable.
If \(e = e_{1}\),
Because
And from Eq. (14), it follows that \(\frac{2}{{9\alpha \beta^{2} }}\left( {\left( {k + 2\alpha \gamma } \right)(1 + \alpha \delta )^{2} - 2\alpha^{3} \beta^{2} } \right) \ge 0\). Hence, Eq. (18) is nonnegative. Namely, \(\left| {\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }}} \right| \ge \left| {\frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}} \right|\). Because of \(\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }} > 0\), \(\frac{{2\left( {1 + \alpha \delta } \right)\sqrt \Delta }}{{9\alpha^{2} \beta^{2} }} \ge \left| {\frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}} \right| \ge \frac{2\alpha }{3} - \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)^{2} }}{{9\alpha^{2} \beta^{2} }}\). As a result, from Eq. (17), \(p - \frac{{ke^{2} }}{2} >\) \(0\) must hold, and \(p > 0\). And from Eq. (13), we have \(e > 0\).
Moreover,
Besides,
Therefore, \(\pi_{P} = \left( {p - \frac{{ke^{2} }}{2}} \right)\frac{{\alpha \left( {1 - \gamma e^{2} } \right) - p}}{{1 + \alpha \left( {\delta - \beta e} \right)}} > 0\). Thus, we have \(e^{*} = e_{1} = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - 2\sqrt \Delta }}{{3\alpha \beta \left( {k + 2\alpha \gamma } \right)}}\), and \(p^{*} = \frac{{\alpha \left( {1 - \gamma e^{*2} } \right)}}{2} + \frac{{ke^{*2} }}{4}\). And we can prove that \(0 < N_{A}^{*} < 1\), and \(0 < N_{C}^{*} < 1\).
From Eq. (19) and Eq. (20), we have \(1 + \alpha \left( {\delta - \beta e^{*} } \right) > 0\), \(\alpha \left( {1 - \gamma e^{*2} } \right) - p > 0\). Combining with Eq. (6), we obtain \(N_{A}^{*} > 0\).
From Eq. (5), we know \(N_{A}^{*} = \alpha N_{C}^{*} - p^{*}\). Because we have proofed \(N_{A}^{*} > 0\), it follows that \(\alpha N_{C}^{*} - p^{*} > 0\). Hence, \(N_{C}^{*} > \frac{{p^{*} }}{\alpha } > 0\).
Further, from Eq. (19), we have \(p^{*} < \alpha \left( {1 - \gamma e^{2} } \right)\). Then, we obtain \(\frac{{1 - \gamma e^{2} + p\left( {\delta - \beta e} \right)}}{{1 + \alpha \left( {\delta - \beta e} \right)}} < \frac{{1 - \gamma e^{2} + \alpha \left( {1 - \gamma e^{2} } \right)\left( {\delta - \beta e} \right)}}{{1 + \alpha \left( {\delta - \beta e} \right)}} = 1 - \gamma < 1\). Hence, from Eq. (7), we prove \(N_{C}^{*} < 1\).
Because
where \(ke^{*2} + 2\alpha \gamma e^{*2} - 2\alpha = \frac{{8\left( {1 + \alpha \delta } \right)\left( {\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta } \right) + 12\alpha^{3} \beta^{2} }}{{9\alpha^{2} \beta^{2} }}\). From Eq. (13), we can find \(\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) > \sqrt \Delta\). Hence, \(ke^{*2} + 2\alpha \gamma e^{*2} - 2\alpha > 0\). And from Eq. (20), we have \(1 + \alpha \left( {\delta - \beta e^{*} } \right) > 0\). Therefore, \(4\left( {1 + \alpha \left( {\delta - \beta e^{*} } \right)} \right) + ke^{*2} + 2\alpha \gamma e^{*2} - 2\alpha > 0\). Based on Eq. (22), we can know \(4\left( {1 + \alpha \left( {\delta - \beta e^{*} } \right)} \right) > 2\alpha \left( {1 - \gamma e^{*2} } \right) - ke^{*2}\). Namely, \(\frac{{2\alpha \left( {1 - \gamma e^{*2} } \right) - ke^{*2} }}{{4\left( {1 + \alpha \left( {\delta - \beta e^{*} } \right)} \right)}} < 1\). From Eq. (21), we prove \(N_{A}^{*} < 1\).
Next, we will verify if the second order conditions hold for \(p^{*}\) and \(e^{*}\). The second-order derivatives of the platform’s profit function are as follows:
From Eq. (20), we have \(1 + \alpha \left( {\delta - \beta e^{*} } \right) > 0\). Thus, \(\frac{{\partial^{2} \pi_{P} }}{{\partial p^{2} }} = - \frac{2}{{1 + \alpha \left( {\delta - \beta e} \right)}} < 0\). And
Since \(\alpha - \frac{{3\left( {k + 2\alpha \gamma } \right)e^{2} }}{2} = \frac{8\sqrt \Delta }{{3\alpha \beta^{2} \left( {k + 2\alpha \gamma } \right)}} > 0\), we have \(\frac{{\partial^{2} \pi_{P} }}{{\partial p^{2} }}\frac{{\partial^{2} \pi_{P} }}{{\partial e^{2} }} - \left( {\frac{{\partial^{2} \pi_{P} }}{\partial e\partial p}} \right)^{2} > 0\). As a result, the Hessian matrix is a negative definite matrix. And the profit function of the platform has the maxima on \(p^{*} = \frac{{\alpha \left( {1 - \gamma e^{*2} } \right)}}{2} + \frac{{ke^{*2} }}{4}\), \(e^{*} = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - 2\sqrt \Delta }}{{3\alpha \beta \left( {k + 2\alpha \gamma } \right)}}\).□
Proof of Proposition 2
Because \(e^{*} = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - 2\sqrt \Delta }}{{3\alpha \beta \left( {k + 2\alpha \gamma } \right)}}\), deriving the above expression with respect to \(\gamma\), we have \(\frac{{\partial e^{*} }}{\partial \gamma } = - \frac{{\alpha^{3} \beta }}{{\left( {k + 2\alpha \gamma } \right)\sqrt \Delta }}\). Since \(\alpha ,\beta ,k,\gamma\) are all positive, thus, \(\frac{{\partial e^{*} }}{\partial \gamma } < 0\).
Besides, \(\frac{{\partial p^{*} }}{\partial \gamma } = - \frac{{\alpha e^{*2} }}{2} - \frac{{\left( {2\alpha \gamma - k} \right)e^{*} }}{2}\frac{{\partial e^{*} }}{\partial \gamma }\). If \(\gamma < \frac{k}{2\alpha }\), it follows that \(2\alpha \gamma - k < 0\). In addition, \(\frac{{\partial e^{*} }}{\partial \gamma } < 0\), we obtain \(\frac{{\left( {2\alpha \gamma - k} \right)e^{*} }}{2}\frac{{\partial e^{*} }}{\partial \gamma } > 0\). Thus, \(- \frac{{\alpha e^{*2} }}{2} - \frac{{\left( {2\alpha \gamma - k} \right)e^{*} }}{2}\frac{{\partial e^{*} }}{\partial \gamma } < 0\). Namely, \(\frac{{\partial p^{*} }}{\partial \gamma } < 0\).□
Proof of Proposition 3
\(\frac{{\partial e^{*} }}{\partial \beta } = \frac{{2\left( {1 + \alpha \delta } \right)\left( {\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta } \right)}}{{3\alpha \beta^{2} \sqrt \Delta }}\). From Eq. (13), we have \(\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta > 0\). Therefore, \(\frac{{\partial e^{*} }}{\partial \beta } > 0\).
Besides, \(\frac{{\partial p^{*} }}{\partial \beta } = \frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \beta }\). If \(\gamma < \frac{k}{2\alpha }\), we have \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2} > 0\). Because \(\frac{{\partial e^{*} }}{\partial \beta } > 0\), we obtain \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \beta } > 0\). Finally, \(\frac{{\partial p^{*} }}{\partial \beta } > 0\). However, if \(\gamma \ge \frac{k}{2\alpha }\), we have \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2} \le 0\). Because \(\frac{{\partial e^{*} }}{\partial \beta } > 0\), and \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \beta } \le 0\)., \(\frac{{\partial p^{*} }}{\partial \beta } \le 0\).□
Proof of Proposition 4
. By \(\frac{{\partial e^{*} }}{\partial \delta } = \frac{{2\left( {\sqrt \Delta - \left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right)} \right)}}{3\beta \sqrt \Delta }\) and Eq. (13), we have \(\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta > 0\). Therefore,\(\frac{{\partial e^{*} }}{\partial \delta } < 0\).
Besides, \(\frac{{\partial p^{*} }}{\partial \delta } = \frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \delta }\). If \(\gamma < \frac{k}{2\alpha }\), we have \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2} > 0\). Because \(\frac{{\partial e^{*} }}{\partial \delta } < 0\), it follows that \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \delta } < 0\), so \(\frac{{\partial p^{*} }}{\partial \delta } < 0\). However, if \(\gamma \ge \frac{k}{2\alpha }\), we have \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2} \le 0\). Since \(\frac{{\partial e^{*} }}{\partial \delta } < 0\), we have \(\frac{{e^{*} \left( {k - 2\alpha \gamma } \right)}}{2}\frac{{\partial e^{*} }}{\partial \delta } \ge 0\), \(\frac{{\partial p^{*} }}{\partial \delta } \ge 0\).□
Proof of Proposition 5
\(\frac{{\partial e^{*} }}{\partial k} = - \frac{{\alpha^{2} \beta }}{{2\left( {k + 2\alpha \gamma } \right)\sqrt \Delta }}\). Because \(\alpha ,\beta ,k,\gamma\) are all positive, \(\frac{{\partial e^{*} }}{\partial k} < 0\).□
Proof of Proposition 6
. By \(\frac{{\partial e^{*} }}{\partial \alpha } = \frac{{2\left( {k + 2\alpha \gamma } \right)\left( {\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta } \right) + \frac{{3k\alpha^{3} \beta^{2} }}{2}}}{{3\alpha^{2} \beta \left( {k + 2\alpha \gamma } \right)\sqrt \Delta }}\), and Eq. (13), we have \(\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta > 0\). Thus, \(2\left( {k + 2\alpha \gamma } \right)\left( {\left( {k + 2\alpha \gamma } \right)\left( {1 + \alpha \delta } \right) - \sqrt \Delta } \right) > 0\). And because the denominator is positive, \(\frac{{\partial e^{*} }}{\partial \alpha } > 0\).□
Proof of Proposition 7
According to the Envelope Theorem, we have
From Eq. (19), it follows that \(p^{*} - \frac{{ke^{*2} }}{2} > 0\) and \(\alpha \left( {1 - \gamma e^{*2} } \right) - p^{*} > 0\). From Eq. (20), we have \(1 + \alpha \left( {\delta - \beta e^{*} } \right) > 0\). As a result, \(\frac{{\partial \pi_{P}^{*} }}{\partial \gamma } < 0\), \(\frac{{\partial \pi_{P}^{*} }}{\partial \alpha } > 0\), \(\frac{{\partial \pi_{P}^{*} }}{\partial \beta } > 0\), \(\frac{{\partial \pi_{P}^{*} }}{\partial k} < 0\), \(\frac{{\partial \pi_{P}^{*} }}{\partial \delta } < 0\).□
Rights and permissions
About this article
Cite this article
Duan, Y., Ge, Y. & Feng, Y. Pricing and personal data collection strategies of online platforms in the face of privacy concerns. Electron Commer Res 22, 539–559 (2022). https://doi.org/10.1007/s10660-020-09439-8
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10660-020-09439-8