The politics of structural choice theory claims that (re-)organizations of government structures are 'structural bargains' between governing parties, seeking to accomplish their policies, but also to accommodate their office preferences and to maximize votes (Moe 1995; Müller/Strøm 1999). Research on coalitional presidentialism has shown that under conditions of multi-partyism, presidents use a wide range of tools to manage coalitions and allocate portfolios (Chaisty et al. 2018). However, structural choices over the allocation and inner organization of ministerial portfolios are also made in dominant-party or single-party systems. Such choices may still express bargains, e.g. between different party wings representing factions of the ruling coalition. Moreover, allocating executive positions is a powerful tool of autocrats to reward allies, co-opt potential competing elites into the ruling coalition or to harness technocratic competence of long-term advisers for regime stability (Buehler/Ayari 2017). In contrast to democracies, however, one may expect less relevance of partisan determinants for explaining structural choices in central government organizations.
We turn to the case of Russia to explore this empirical puzzle by analyzing the first longitudinal quantitative analysis of all formal structural changes of Russian federal government ministries between 2000 and 2014. Previous research on autocracies focused on bureaucratic appointments to investigate trade-offs and effects of the contrasting selection criteria of loyalty versus competence (Li/Zhou 2005; Reuter/Robertson 2012). This is also due to the fact that informal networks and patronage ties are usually thought to have prevalence over formal structures. With the institutional turn in comparative authoritarianism (Frye 2012; Pepinsky 2014), the origins, functions, and effects of institutions and organizations have taken center stage. But due to the inherent secrecy in these regimes, data on structural characteristics of presidential governments is sparse and thus usually does not allow for quantitative analysis.
