Pension Plan Allocation to Real Estate when Plan

  • Slides: 14
Download presentation
Pension Plan Allocation to Real Estate when Plan Trustees have Reputational Utility Kiat-Ying Seah

Pension Plan Allocation to Real Estate when Plan Trustees have Reputational Utility Kiat-Ying Seah and James D. Shilling National University of Singapore De. Paul University, Chicago ERES 2010 Milan, Italy

Objectives p Tries to explain why institutions invest very little in real estate. n

Objectives p Tries to explain why institutions invest very little in real estate. n n An entropy model where institutions care about what their target returns are Conformity matters Persistent portfolio allocation Implication: allocation is based on a power utility not on a mean-variance variety

Motivation p p Reality: institutions invest 2. 5% to 4% of total assets in

Motivation p p Reality: institutions invest 2. 5% to 4% of total assets in real estate Normative studies: n n n p 15 -20% (Fogler, 1984) 43% (Webb and Rubens, 1987) 19 -28% (Giliberto, 1993) Modelling idea: - Pension plans are fiduciaries and care about how others assess their performance: include “Reputation” in utility function.

Pension Trustee’s Objective function p Measure reputation by an entropy function p Objective of

Pension Trustee’s Objective function p Measure reputation by an entropy function p Objective of each plan trustee: maximize reputation subject to a shortfall constraint

Allocation is a function of beliefchoice p Optimal belief choice: multinomial logit probabilistic choice

Allocation is a function of beliefchoice p Optimal belief choice: multinomial logit probabilistic choice function n n Trustees will skew their portfolio toward assets with higher returns. Prob of choosing a target that deviates from group mean is low – conforming behavior.

Given beliefs, solve for portfolio 1. Power utility form, risk aversion is endogenous 2.

Given beliefs, solve for portfolio 1. Power utility form, risk aversion is endogenous 2. What matters? • Target returns, W 0 Z • Surplus returns ST Initial Funding Ratio matters

Data and Results Data are obtained from CRSP/COMPUSTAT p 1990 -2004. p Summary Statistics:

Data and Results Data are obtained from CRSP/COMPUSTAT p 1990 -2004. p Summary Statistics: p

Summary Statistics

Summary Statistics

Portfolio Simulations 1. SRMP looks at total portfolio variance 2. Entropic cares about the

Portfolio Simulations 1. SRMP looks at total portfolio variance 2. Entropic cares about the entire distribution

Empirical Evidence p Use Sharpe’s (1992) “style” methodology: n n Regress pension surplus returns

Empirical Evidence p Use Sharpe’s (1992) “style” methodology: n n Regress pension surplus returns on six benchmark returns. Collect R-square statistic for each pension plan. Because R-square varies from 0 to 1, transform this variable using the logistic transformation = “STYLE” variable Run the following regression: Style = F (Target surplus return, Conformity, Initial Funding Ratio)

Conclusion Empirical results are the same when we include firm fixed effects. p Reputational

Conclusion Empirical results are the same when we include firm fixed effects. p Reputational utility causes institutions to p n n n Skew portfolio away from real estate to achieve minimum target rate of return. Achieving minimum target rate of return requires that pension trustees be conformists. Explains herding behavior.