(with Guillermo Diaz and Christos Mavridis)
We evaluate the effect of an institutional provision designed to increase accountability of local officials, and we show that its implementation can lead to a distribution of power within the legislature which is not consistent with voters' true preferences.
The cause of this inconsistency is the ballot design which asymmetrically affects the officials listed on it. We analyze the case of the Lima's 2013 city legislature recall referendum and show that the design of the referendum ballot had adverse and significant effects on the composition of the Lima's city legislature. We also show that the election results with more ``neutral'' ballot designs would have been significantly different, and the composition of the council would have been more representative of voters' true preferences. More specifically, we use our results to simulate the outcome of the election with a random order of legislators. Even though the voters' fatigue is still present, it affects all parties equally, obtaining a more faithful representation of the voters' preferences. With this ``random ballot'', the provision would keep politicians accountable.