RESEARCH

Book

Image of Book with announcement of Winner Charles H Levine Memorial Book Prize

When Democracies Deliver: Governance Reform in Latin America. Cambridge University Press, 2019.

Why do governance reforms in developing democracies so often fail, and when might they succeed? When Democracies Deliver offers a dynamic framework for assessing the effectiveness and durability of policy change. Drawing on detailed analyses of public sector reforms in Brazil and Argentina, this book challenges conventional wisdom to reveal that incremental changes sequenced over time prove more effective in promoting accountability, increasing transparency, and strengthening institutions than comprehensive overhauls pushed through by political will. Developing an innovative theory that integrates cognitive-psychological insights about decision making with research on institutional change, Katherine Bersch shows how political and organizational factors can shape reform strategies and information processing. Through extensive interviews and field research, Bersch traces how two competing strategies have determined the different trajectories of institutions responsible for government contracting in health care and transportation. When Democracies Deliver offers a fresh insight on the perils of powering and the benefits of gradual reform.

Winner of the Charles H. Levine Memorial Book Prize from the International Political Science Association’s Research Committee on the Structure of Governance, the Donna Lee Van Cott Award for the Best Book on Latin American Politics and Institutions from the Political Institutions and Processes Section of the Latin American Studies Association, and the SPAR Award for the Best Book Published in Public Administration in 2019 by the American Society for Public Administration.

An article that advances the main theory of my book project can be found in the January 2016 issue of Comparative Politics.

The Stanford Governance Project

My current research with Francis Fukuyama on the Stanford Governance Project extends my interests in state capabilities cross-nationally. Drawing on surveys of public sector employees as well as big data on governments, survey experiments, and qualitative methods, we examine the determinants of bureaucratic performance and consequences for public service delivery in the developing world. In 2018, I conducted a survey of over 3,200 federal public sector workers in Brazil, collaborating with one of Brazil’s most respected governmental research institutions (IPEA). We then tested the hypothesis outlined in Fukuyama’s influential “What is Governance?” article, wherein he argues that the relationship between bureaucratic discretion and government quality follows an inverted U. Our findings support this Goldilocks hypothesis: too little discretion reduces quality, but so does too much. Likewise, too much or too little autonomy among political actors reduces control of corruption. Results also suggest that agency capacity plays an important role in moderating these relationships.

The Global Survey of Public Servants

Extending my work with the Governance Project, I co-founded the Global Public Servant Survey (GPSS), an international research initiative to better understand how public services work and to help governments manage public servants.  In 2018, I was invited to the World Bank to discuss my findings from the Brazil survey. I then helped initiate a collaboration among the World Bank, Stanford, University College London, and Davidson dedicated to developing a common, cross-national survey. Our team forges partnerships with host governments and local affiliate research teams, with the goal of producing a cross-national, longitudinal dataset of public servants survey data.

State Capacity

The Stanford Governance Project builds on my ongoing research that analyzes the variation of bureaucratic capacity and political autonomy within and across national governments and evaluates the implications for effective governance, democratic performance, and public policy. The findings of this research coauthored with Matthew Taylor and Sérgio Praça are the subject of an article in Governance (2017) and a chapter in Miguel Centeno, Atul Kohli, and Deborah Yashar’s edited volume with Cambridge, State Building in the Developing World (2017). This study uses Bayesian latent variable analysis, an innovative methodology that I elaborated and which first appeared in my coauthored article, “Measuring Governance: Implications of Conceptual Choices,” published in European Journal of Development Research (2014). Drawing on a novel dataset that includes individual-level data of the careers and political affiliations of 326,000 civil servants, we produced objective measures of state capacity and political autonomy for 95 federal agencies in Brazil.