Economics of Ransomware: Risk Interdependence and Large-Scale Attacks

• 成果类型: Article
• 署名作者: August, Terrence; Dao, Duy; Niculescu, Marius Florin
• 署名单位: University of California System; University of California San Diego; University of Calgary; University System of Georgia; Georgia Institute of Technology
• 刊物名称: MANAGEMENT SCIENCE
• ISSN/ISSBN: 0025-1909
• DOI: 10.1287/mnsc.2022.4300
• 发表日期: 2022
• 页码: 8979-9002
• 关键词: software security; software patching; Ransomware; cyber attacks; network externalities

摘要

Recently, the development of ransomware strains and changes in the marketplace for malware have greatly reduced the entry barrier for attackers to conduct large-scale ransomware attacks. In this paper, we examine how this mode of cyberattack impacts software vendors and consumer behavior. When victims face an added option to mitigate losses via a ransom payment, both the equilibrium market size and the vendor's profit under optimal pricing can actually increase in the ransom demand. Profit can also increase in the scale of residual losses following a ransom payment (which reflect the trustworthiness of the ransomware operator). We show that for intermediate levels of risk, the vendor restricts software adoption by substantially hiking up price. This lies in stark contrast to outcomes in a benchmark case involving traditional malware (non-ransomware) where the vendor decreases price as security risk increases. Social welfare is higher under ransomware compared with the benchmark in both sufficiently low- and high-risk settings. However, for intermediate risk, it is better from a social standpoint if consumers do not have an option to pay ransom. We also show that the expected ransom paid is nonmonotone in risk, increasing when risk is moderate despite a decreasing ransom-paying population. For ransomware attacks on other vectors (beyond patchable vulnerabilities), there can still be an incentive to hike price. However, market size and profits instead weakly decrease in the ransom amount. When studying a generalized model that includes both traditional and ransomware attacks, our results remain robust to a wide range of scenarios, including threat landscapes where ransomware has only a small presence.