Sponsor Bias in Pension Fund Administrative Expenses: The Brazilian Experience.

Autorda Cunha, Claudio Marcio Pereira
CargoReport

Introduction

The importance of pension funds grows as populations age. Pension values depend directly on the net return obtained on pension fund assets. Administrative expenses deducted from gross return may play a significant role in net return. In a simulation performed by Bikker and De Dreu (2009), if annual administrative expenses increase from 1% to 2% of total fund assets, retirement income may drop around 25% (assuming 5% of annual real rate of return).

The literature concerned with pension fund administrative expenses has two main goals. One goal is to identify potential scale economies, as in Caswell (1976). If there are economies of scale, merging pension funds may be beneficial to participants. Consequently it is interesting to find out if there is an optimal pension fund size, as in Bikker (2017). The other goal related to pension fund administrative expenses is concerned with the factors that drive them, as in Bikker, Steenbeek and Torracchi (2012), that investigate the effect of plan characteristics. In this study we develop the second approach, dissecting previous findings.

In most developed countries, the institutional structure of pension systems are organized upon 3 pillars. The first pillar comprises a public pension scheme, usually mandatory, financed on a pay-asyou-go (PAYG) basis. Because it usually pays a defined benefit and may be complemented by taxes, the effect of administrative expenses to participants is limited. Most studies on administrative expenses focus on the second pillar, which comprises closed pension funds. These are collective private organizations, where participants share a professional link to the sponsor, an organization that may be public, private or professional. Closed pension funds are financed solely by contributions from participants and sponsors, implying that contributions and benefits are directly affected by fund performance, thus, by administrative expenses. The third pillar are tax-deferred individual accounts provided by financial institutions, compensated by fees, charged usually as a percentage of account balance. There is freedom to move individual accounts from one institution to another (portability). Thus, administrative expenses are less relevant, because participants do not necessarily benefit from low expenses and they can more easily avoid high expenses (Bikker, Steenbeek, & Torracchi, 2012; Bikker, 2017).

In Brazil, the basic difference from the system described above is that the public pension scheme is divided into a General Social Security System (in Portuguese, Regime Geral de Previdencia Social - RGPS), for private sector workers, and a pension regime for government workers (called, in Portuguese, Regimes Proprios de Previdencia Social - RPPS) (Associacao Brasileira das Entidades Fechadas de Previdencia Complementar [ABRAPP], 2014).

However, the focus of present study is not on public pension schemes, but on administrative expenses of Brazilian complementary sponsored closed pension funds, known by the abbreviation EFPC (Entidades Fechadas de Previdencia Complementar, in Portuguese), which are private entities that correspond to the second pillar in the system described for most developed countries.

EFPC are directed at medium and high income workers to preserve their living standards after retirement, since the GSSS (or, in Portuguese, Regime Geral de Previdencia Social [RGPS]) has a ceiling for the value of pensions, and has been oriented to play a distributive role. Sponsors may be state-owned companies (public sponsor), private companies (private sponsor), or labor unions and professional associations (group sponsor) (ABRAPP, 2014).

EFPC represent a large share of savings in Brazil, being responsible for a large fraction of retirees' and beneficiaries' incomes. By the end of 2015 the summed assets of EFPC added up to R$ 721 billion (US$ 182 billion), representing 12% of Brazilian gross domestic product (GDP). In 2014, for 28% of EPFC, yearly administrative expenses exceeded 1% of total assets, meaning a reduction of more than 1% on yearly net returns (Superintendencia Nacional de Previdencia Complementar [PREVIC], n.d.).

Bikker et al. (2012), for Australia, Canada, the Netherlands, and the United States, and Abi-Ramia, Boueri and Sachsida (2015), for Brazil, have reported a sponsor bias in administrative expenses of closed pension funds. Both studies show evidence that funds with public sponsors (either state-owned companies or local governments) present greater administrative expenses, after controlling for assets, number of participants and fund characteristics. However, neither of them investigated the reason for this sponsor bias. The main objective of the present study is to investigate possible causes for the greater administrative expenses for public sponsored pension funds (sponsor bias) in Brazil.

Hypotheses

We test three hypotheses that may explain the sponsor bias. First, we hypothesize that public sponsored pension funds are located in cities with higher costs of living, which imposes higher prices to production factors, mainly labor. Indeed, almost half of public sponsored pension funds are related to the federal government, which implies that most of the employees are located either in Brasilia (the capital of Brazil), or in Rio de Janeiro (the former capital of Brazil, that still concentrates some federal activities). These metropolitan regions, along with Sao Paulo, historically present the greatest costs of living in Brazil (Almeida & Azzoni, 2016). We evaluate the influence of location on total administrative expenses using dummies for location.

Our second hypothesis is that public sponsored EFPC insource portfolio management activities. A way to outsource portfolio management is to allocate assets to investment funds. In this case, the cost of portfolio management may be deduced from the net return of the asset, instead of being accounted for as administrative expenses. Thus, insourcing portfolio management may imply higher administrative expenses when compared with similar pension funds that outsource this activity. There are at least three reasons for public sponsored EFPC to insource portfolio management more than other EFPC. First, public sponsored EFPC have more assets than private or group sponsored pension funds, meaning that they benefit more from insourcing activities due to the economies of scale. In 2015 there were 76 public sponsored EFPC, representing 28% of the total number, and 64% of summed assets of all EFPC (PREVIC, 2015). Second, insourcing portfolio management allows activism by pension funds in equity investment, that can improve firm valuation, as reported by Giannetti and Laeven (2009). But, again, pension funds must be large to benefit from activism, because it is necessary to own large stakes of companies to influence their administration. And third, discretionary power in portfolio allocation may also be used politically, as reported by Bradley, Pantzalis and Yuan (2016) , and public pension funds would be more exposed to political interference.

Finally, we hypothesize that public sponsors may have discretionary power over administrative expenses, directing fund resources to support political activities. In Brazil, Lazzarini, Musacchio, Bandeira-de-Mello and Marcon (2015) report that the firms that donate to winning candidates are more likely to receive funding in the form of loans from the state owned National Development Bank (Banco Nacional de Desenvolvimento Economico e Social - BNDES). Carvalho (2014) provide evidence that BNDES may have acted to induce employment expansion in politically attractive regions immediately before competitive elections. Thus, we test if administrative expenses of public sponsored EFPC increase in electoral years, which would indicate the use of fund resources to influence elections.

Pension Funds Administrative Expenses

The modelling of pension fund administrative expenses starts with Caswell (1976), who was primarily concerned with the existence of economies of scale in pension plan administration. With significant economies of scale, merger of small plans present potential for cost savings. Caswell (1976) recognizes two major products that demand administrative expenses: services to participants and investment activities (portfolio management). He considers the total number of participants and total assets as adequate output proxies, respectively. However, the analysis only took the number of total participants as an output proxy, along with control variables, arguing that it would be inadequate to consider both output proxies, due to strong correlation between them.

Mitchell and Andrews (1981) generalize Caswell's methodology. They postulate a Cobb-Douglas cost function, to consider both outputs in the same analysis, modelling administrative expenses as represented in equation (1), where AdmExp is total administrative expenses, Pop is the number of total participants, Assets is total assets, P refers to input prices, [epsilon] is an error term, [alpha], [[beta].sub.1], [[beta].sub.2], and [phi] are parameters of the model, and i is an index that identifies pension plans.

In([AdmExp.sub.i]) = [alpha] + [[beta].sub.1]ln([Pop.sub.i]) + [[beta].sub.2]ln([Assets.ssub.i]) + [phi] ln([P.sub.i]) + [[epsilon].sub.i] (1)

In their final model, represented in equation (2), Mitchell and Andrews (1981) add control variables (the vector X), with a corresponding vector of parameters ([gamma]), and omit input prices (P), arguing that prices do not vary systematically across plans, but acknowledge that this omission would be serious in a time series study. However, if plan headquarters are placed in different locations and labor costs are significantly different between locations, a proxy for input prices should be considered. If location is correlated with any of these control variables, parameters estimates from linear regressions will probably be...

Para continuar a ler

PEÇA SUA AVALIAÇÃO

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT