Modeling Liquidity and Income in Modern Portfolios Todd E Petzel, CIO Offit Capital Advisors QWAFAFEW New York March 24, 2009
Outline of Presentation – Traditional Models and Assumptions – Incorporating Liquidity and Income Theoretically – Practical Issues and Approaches
Traditional Approaches –Linear or non-linear optimization in return space –Monte Carlo simulation in return space –“Total Return” spending rules Major implicit assumption: portfolio adjustments are frictionless and costless
Optimizers have multiple problems Thousands of data points; one history Corner solutions are the norm “Solutions” more likely to reflect constraints than truth
Monte Carlo is supposed to cure these issues Thousands of simulations, but based on same history Distribution of outcomes versus a single expected characterization
Monte Carlo approach still has severe issues Covariance assumptions are subject to abrupt changes Path dependency is fairly rudimentary Still backward looking
December 31, 2008
June 30, 2008
Total Return Spending Rules The exception rather than the rule 40 years ago Assumes sufficient liquidity to create payments from portfolio and to rebalance Ignores actual operations side of enterprise and covenants
Private Equity Simulation Rules First cousin to Monte Carlo portfolio analysis Used to plan transition to “long-term” portfolio containing illiquid partnerships Usual conclusion: Over allocate to illiquid partnerships in order to reach goal Keep money in equities while waiting for calls
Major Unstated Assumptions Bull markets provide early distributions and funding sources for following calls There will always be enough liquid securities to sell when capital calls appear Simulations based on a decidedly bull market history
Reality in 2008 PE obligations slowed down, but still remain dollar liabilities to the investor Intended source of funding hammered by bear market Liquid securities have been sold down to meet regular spending and capital calls Major institutions borrow to pay bills
Where do Liquidity and Income Fit In? In the traditional approaches there is no difference between liquid and illiquid investments, or between income and total return Recommendations for illiquid private investments are usually only bounded by initial constraints
How to Improve the Models Don’t maximize wealth, maximize utility U = f(W, L, I) [Wealth, Liquidity, Income] Downward sloping marginal utility of all factors Upward sloping transactions costs associated with less income or liquidity Higher opportunity costs of more income and liquidity
Conceptually This Isn’t Too Difficult Problems arise in execution Do organizations understand their marginal utilities of liquidity and income in good times? Very similar problem to estimating the marginal utility of storage between times of full inventories and shortages.
Practical Approaches Throw away your total return spending rule Integrate the operational budget and investment processes Understand how much cash you’ll need in the near term
Practical Approaches II Split the portfolio into two components: Sleep well at night money Long-term portfolio Try to cover cash needs with income producing assets If that is not possible, decrease illiquid assets to lower impact of asset sales
Practical Approaches III Forecast future capital calls Set aside “sleep well at night money” for these liabilities extending some period Do not over allocate to partnerships to try to build up positions quickly
Conclusions Inability to properly model income or liquidity benefits skewed portfolio construction toward higher risks Too many institutions are revisiting these topics now after suffering permanent losses Ad hoc rules are better than inadequate models