The last two decades have revealed the vulnerability of privately owned "critical infrastructure"-the power grid, pipelines, financial networks, and other vital systems-to cyberattack. The central U.S. response to this challenge has been a series of sectoral "partnerships" with private owner-operators of critical infrastructure, involving varying degrees of regulation. Qualitative analysis based on in-depth interviews with over 40 policymakers and senior private sector managers, as well as public documents, reveals considerable variation in how well this approach has worked in practice. The main predictors of policy success appear to be (a) the nature of the cyber threat to firms' operations and (b) regulatory pressure on firms. However, other factors-such as the nature of intra-industry competition-also affect how well the current regime works in specific sectors. Our findings have implications for public administration on civilian cybersecurity, as well as ramifications for regulation in other policy domains.
Evidence for Practice• Collaboration between business and government in cybersecurity is distinct from conventional publicprivate partnerships designed to address capital markets failures, in that it must be highly flexible and adaptive. • Cybersecurity policies should be tailored to critical infrastructure sectors or subsectors, to take into account the nature of industry competition, the size and complexity of the sector, and longstanding relationships between business and the government in the sector. • Government agencies that possess a strong historical relationship to their assigned sector, expertise in cyber, and resources to help firms are better able than other lead agencies to build an effective cybersecurity partnership with industry. • Irrespective of sector, collaboration between the government and private owner-operators of critical infrastructure requires a high level of trust, often built through personal relationships and then reinforced through iterated interactions.C ritical infrastructure refers to the systems that undergird modern society: the power grid that provides electricity to businesses and households, financial networks that allow the market economy to function, water and sewerage systems, and the like (Alcaraz and Zeadally 2015; CIPA 2001; DHS 2019). Because ordinary operations in these sectors are increasingly digitalized, and because the hardware and software components of these systems often have exploitable features (NIST 2019), much of this infrastructure is susceptible to cyberattack (Clinton and Perera 2016;Johnson 2015). The increasing connectivity of critical infrastructures to other networks, including the internet, has exacerbated this vulnerability (inter alia Clinton 2016; Speake 2015; Johnson 2015).Not only do attacks threaten targeted firms and sectors themselves, but they could also potentially trigger cascading failures (