Sponsor Bias in Pension Fund Administrative Expenses: The Brazilian Experience

Abstract Previous literature has reported that pension funds sponsored by public organizations present greater administrative expenses when compared to similar pension funds sponsored by private organizations. We investigate this sponsor bias, hypothesizing that it may originate from the omission of relevant control variables, specifically variables for location of headquarters and the level of outsourced services. We test this hypothesis by linear regression in the cross-section of 164 Brazilian closed pension funds, using annual data from 2010 to 2014. We find that these control variables partly explain the sponsor bias, especially for medium-size pension funds, and when the sponsor is an organization related to a state or municipal government. We also hypothesize that political bias may increase administrative expenses of public sponsor pension funds, especially in election years. We test this hypothesis by panel regression using a fixed effects method and did not find statistically significant changes in administrative expenses in election years. Our findings do not support the hypothesis of political bias in administrative expenses of Brazilian closed pension funds. On the contrary, we present evidence that the sponsor bias may be driven by characteristics of the pension funds omitted in previous literature.