Reading 13: Principles of Asset Allocation Flashcards
Personal Residence / Private Company
You want to incorporate these into MVO to model risk and return (incorporating these large assets can allow investor to take more risk). You can also ignore them, if investor is not likely to ever sell them. You can include them in the modelling by using data on public funds but realizing that such data will not reflect the specific characteristics of the client’s positions.
Single Periods
MVO can address taxes in a single period model (rebalancing not necessary) and MCS is not necessary. MCS becomes useful in dealing with a multi-period framework, where the analysis of taxes and rebalancing becomes much more mathematically challenging otherwise (taking into consideration cash flows). MCS compliments MVO, not replaces. MCS can be used to incorporate investor preferences into AA.
Asset Class Vs Factor Based Allocation
Not all factors are replicable, but the ones that are can be obtained with a position that is long the desired factor (e.g. growth) and short the undesired factor (e.g., value). Neither asset class–based allocation nor factor-based allocation is superior. It is a matter of which method better resonates with how the manager looks at investing, not the kind of client.
Integrated Asset-Liability Approach
Used typically by banks (as their liabilities change). It includes joint optimisation and focuses on both assets and liabilities together with a continuous feedback loop between the two, requiring a multi period approach. The other two approaches in liability relative asset allocation (Surplus Optimisation & Two-Portfolio Approach) take liabilities as a given (in place before asset allocation decisions made) and only focus on managing the assets.
Active Risk
Overall market risk is relevant in an asset allocation setting and active risk is relevant in an asset allocation implementation setting.
The Endowment Model
(or Yale model) allows for higher allocation to alternative investments, than recommended under mean-variance optimization (MVO). The investor should select managers with significant exposure to these alternative asset classes. Looks to earn liquidity premiums, suggests a lack of informationally efficient markets.
Goals Based: Insto Vs Individuals
Return for individuals is measured on minimum expectations but for insto’s it is the mathematical expectations (weighted expected returns of components). Insto’s have one goal usually, and individuals have multiple. Risk measure for individuals is missing the goal. Risk measure for insto’s is volatility of return on surplus.
Illiquid Assets
Indexes that track less liquid asset classes (i.e. direct real estate, infrastructure, private equity) are typically not investable as a passive alternative to active management. Therefore, they are difficult to include in MVO. Illiquid assets themselves cannot be readily diversified so the investment vehicles may not represent characteristics of the asset class (that is completely diversified). They can be included (using highly diversified asset classes or the risk characteristics of the asset instead of asset class) or excluded in MVO.
Two Portfolio Approach: Liability Relative AA
The hedging portfolio could be constructed using various techniques such as cash flow matching, duration matching, and immunization. If the funding ratio is less than 1, then it becomes difficult to create a hedging portfolio that completely hedges the liabilities. The composition of the liabilities is already in place when the asset allocation decisions are made, so the two decisions are made independently.
MVO
Results in AA that are heavily concentrated in subsets of asset classes and use historical inputs. Black-Letterman and reverse optimisation model would most likely be less concentrated in specific subsets. Model outputs tend to be highly sensitive to inputs.
Rebalancing Corridors
Higher risk assets in practise should have wider corridors to avoid frequent rebalancing and transaction costs. In theory they would warrant narrow corridors. Less liquid assets should have wider corridors to avoid frequent rebalancing costs. Taxables investors should have wider corridors to avoid capital gains tax.
Risk Parity AA
Notion that each asset class should contribute equally to total portfolio risk. If bonds for example have lower risk, then they should be overweighted relative to other asset classes.
Factor Based AA
The factors generally have low correlation with the market and with themselves. This is because the factors typically short the underperforming attribute (large stocks, growth stocks low B/M) and long the better performing attribute. The factors used are similar to the fundamental or structural factors in multi-factor models. They are called zero dollar portfolios.
Reverse Optimisation
The inputs are the asset allocation weights typically of the global market portfolio (assumed optimal). The asset allocation weights for MVO are the outputs, with inputs being expected returns, covariances, forecasted with historical data and risk aversion coefficient. The two methods result in different AA. For reverse, the inputs are market cap weights, covariances, and risk aversion coefficient and the output is the expected optimised return (long run capital return, linked to CAPM).
MVO Criticisms
Garbage in garbage out; concentrated asset class allocations; does not consider skewness and kurtosis; risk diversification over asset classes with same risk factors is possible; ignores liabilities; single period framework not considering interim cash flows or serial correlations.
