How to separate the office from the officeholder is one of the most difficult questions in the empirical study of institutions and leadership. We argue that provided there is an indicator for the overall individual influence among members of the political elite and there is sufficient variability among individuals taking the same office, being promoted and demoted into different offices over time, we can separate latent individual and institutional components of influence at an aggregate, regime level. Our latent variable model thus provides a new tool to measure the degree of regime deinstitutionalisation.Using expert surveys that assess the ranking of the top political actors in Russia from 1994-2011, and restricting personal effects to those that are constant over time, we find that on average office dominates individual by the order of two. We discuss regime deinstitutionalisation in comparative perspective; demonstrate the generalisability by analysing Ukraine; account for patronage networks.Key Words: leadership, deinstitutionalisation, latent factor model, Russian and PostSoviet politics Does the political office that an individual occupies primarily determine one's overall political influence, or can the same office be strengthened or weakened depending on the officeholder? Recently, there has been a surge of scholarship that examined with new rigour whether leaders matter (e.g., Dewan and Myatt, 2008;Goemans and Chiozza, 2009;Horowitz and Stam, forthcoming;Jones and Olken, 2009). Nonetheless, it remains extremely difficult to attribute causality to particular officeholders, discern the effects of leaders from other factors, or disentangle the office from the individual officeholder (Ahlquist and Levi, 2011;Shamir, 2012;Sheffer, 1993). This paper directly confronts the empirical difficulty of separating the effects of leaders on policy, thereby addressing a centuries-old conundrum in political science. We do this through a novel measurement tool for the separation of different sources -or components -personal (officeholder) and institutional (office) -of the individual's influence. This approach largely permits us to side-step assumptions about the relative effects of individual traits, followers and contextual factors, and to identify the effects of leaders and institutions in a non-experimental setting. The expert survey we rely on provides one observed variable measuring the perceived policy influence for the most important actors within a political regime over time and thus the data itself cannot, at first sight, provide information on the relative importance of individual and office. 1 However, because over time different actors can hold the same office and the same actor can hold different offices, we can capture the corresponding changes in influence scores and use this variation to gauge the relative importance of the person and the office. We therefore measure, at an individual level, the relative attribution of influence to office and individual, and, at an aggregate level, the 1 An onl...