Andrea Reed, Cristian Tiu and Uzi Yoeli
Volume 9, Number 1, First Quarter 2011
The process of risk management for institutional investors faces two challenges. First, since most institutions are decentralized in contrast to being direct investors in assets, it is difficult to separate the risks of the assets in the portfolio from the risks generated by the investment decisions by fund management. To address this issue, we propose a risk measurement methodology which calculates the risk contributions of individual securities and investment decisions simultaneously. This decomposition is applicable to any decentralized investor as long as its relevant risk measurement statistic can be additively decomposed. Second, statistics used to measure risk may not coincide with institution-specific investment risks, in the sense that the utility employed in asset allocation may be unrelated to the risk measure utilized. For example, an institution may do mean-variance asset allocation, but inconsistently measure the risk of the portfolio using Value at Risk. We apply this methodology to a particular type of decentralized investor, specifically, endowment funds where the relevant risk statistic is the downside risk of returns relative to actual payout levels, plus inflation. We show how downside risk can be decomposed and apply our simultaneous downside risk decomposition empirically on a sample of U.S. endowment funds. We find that an endowment’s asset allocation to U.S. Equity, consistent with having the largest weight in the average endowment portfolio, generates almost half of the total endowment returns but almost 100% of the total portfolio downside risk. We further find that tactical allocations (or timing) have economically small contributions to both returns and risk. Finally, we find that the allocations to U.S. Fixed Income and to Hedge Funds as well as active investment decisions (except for tactical) contribute positively to returns, while reducing portfolio downside risk.