Does Asset Allocation Influence Portfolio Performance?: Evidence From University Endowment Funds

 

Keith C. Brown

Lorenzo Garlappi

Cristian Tiu

 

 

August 2008

 

 

Abstract

 

In an efficient market, the asset allocation decision is the sole determinant of a portfolio's expected return. Using a unique data set for university endowment funds, we test this hypothesis and find the evidence to be strongly contradictory. We show that although endowments differ substantially in their capital commitments to various asset classes, the volatility and the associated policy portfolio returns are remarkably similar across the sample, indicating that asset allocation is unrelated to returns in the cross section. Moreover, while the risk-adjusted performance of the average endowment is not reliably different from zero, more actively managed funds generate statistically and economically significant annual alphas that are three to six percent greater than those for more passive endowments. This finding is consistent with endowment managers attempting to exploit their security selection ability by over-weighting asset classes in which they appear to have superior active management skills. Finally, we demonstrate that the average endowment in our sample under-utilizes active management and that its out-of-sample performance would improve by taking more active risk while keeping its policy allocation unaltered. Contrary to both efficient market theory and prevailing industry beliefs, we find that asset allocation is not related to portfolio returns in the cross section but does appear to indirectly influence risk-adjusted performance.

 

Download this paper (PDF format

 

Return to Keith Brown’s Working Paper Page