Bureaucratic turnover and politicisation in the Chilean state

An extensive literature has argued that a professional, meritocratic civil service that is shielded from extensive politicisation is conducive to better political and policy outcomes. The measurement of politicisation of the civil service, however, has been mostly approached through cross-national perception-based expert surveys (e.g. on the extent of meritocratic hiring), which allow for scant differentiation within different agencies or sectors of the bureaucracy and also between different levels of it. The latter is relevant inasmuch as it has been suggested that patronage can be used for very different ends, such as controlling policymaking at the higher echelons of the bureaucracy – a strategy typical of more developed civil services – and rewarding supporters or sustaining party machines at the lower end. In this paper, we contribute to overcoming these limitations by studying the politicisation of the Chilean civil service through a newly assembled dataset that is equivalent to an almost-complete personnel registry of its central government bureaucracies. Covering over 300,000 individuals and over 15 million data points across more than 200 agencies, the panel allows us to analyse the turnover dynamics of Chile’s central government over 14 years (2006 – April 2020), including three presidential elections and their aftermath. As a country with a relatively well-developed bureaucracy for Latin American standards (IADB 2013), and yet as one of the newer and less developed countries of the OECD, Chile provides a test study of bureaucratic turnover dynamics and politicization outside the group of rich, highly stable democracies. Preliminary findings suggest that yearly turnover is not dissimilar to that reported for the U.S. federal civil service (around 12%), but turnover varies highly across agencies, thus highlighting the need for more nuanced measures of bureaucratic dynamics. We also find that, after elections, turnover increases not only for top civil servants but across the board, though decreasingly so as one moves down the bureaucratic hierarchy. The structure of turnover also has a clear temporal pattern whereby turnover is highest for top civil servants in the first year after the election, while turnover for middle-ranking officials peaks in the second year. The pattern is consistent with that found in more developed civil services where control of policymaking is the key driver of turnover. However, the finding that turnover percolates down the hierarchy over time suggests a less than fully Weberian civil service. Overall, our results help us to identify both the Chilean state's "islands of excellence" and its zones of more extended political patronage. They also lend support to Gingerich’s (2013) call for disaggregated measures of national bureaucracies to avoid the “level of analysis” problem when treating the state as a causal factor in political science.