Every frontier in the mean-variance optimization module was computed for somebody. The optimiser does not know who, and that is the gap this section closes, because an efficient portfolio for the wrong investor is not efficient in any sense that matters.
What are the stages of the portfolio management process?
The industry's standard frame is a three-stage loop, not a project with an end date.
Planning produces the investment policy statement. It captures the client's return requirement, risk tolerance, and the five constraints (liquidity, time horizon, tax, legal and regulatory, and unique circumstances), then converts them into a strategic asset allocation and a benchmark. PW1.01 owns the private-client version of this document in full, including the ability-versus-willingness distinction and the drafting conventions; AA1.05 owns the institutional version. This node owns only what happens once the return and risk targets exist as numbers.
Execution turns the policy portfolio into positions: manager or vehicle selection, portfolio construction inside each sleeve, and the trading itself. Two costs arise here that the frontier never shows, implementation shortfall against the decision price, and the tracking error a real vehicle carries against the asset-class index the CMA table assumed.
Feedback is monitoring, rebalancing and performance evaluation, and it is where the loop closes back into planning. Rebalancing policy is AA1.02's subject. Performance measurement, attribution and the GIPS presentation standards are AA1.03's. When feedback shows the client's circumstances have moved rather than the market, the policy statement is rewritten and the loop starts again.
The reason the loop matters commercially is that the second and third stages consume almost all of the fee and almost none of the analytical attention. A firm that redoes its optimisation quarterly and never measures its own implementation shortfall has automated the cheap part.
Advertisement
How do the constraints differ by investor type?
The five constraints take systematically different values by investor type, and knowing the pattern is faster than deriving it case by case.
| Investor | Return objective | Risk tolerance | Liquidity need | Time horizon | The binding constraint in practice |
|---|---|---|---|---|---|
| Individual, accumulation phase | Growth, real | Moderate to high, and behaviourally fragile | Low, beyond an emergency reserve | Long, 20 to 40 years | Willingness, not ability; the client sells at the bottom |
| Individual, retired | Income plus inflation protection | Low to moderate | High and recurring | Medium, and uncertain by construction | Longevity risk and the sequence of returns |
| Defined-benefit pension | Fund the liability, so the discount rate is the target | Depends on funded ratio and sponsor strength | Predictable, actuarially known | Long, but liability-matched by duration | The liability, which is a bond the plan is short |
| Endowment or foundation | Spending rate plus inflation, perpetually | High, given the horizon | Low, roughly the annual spending draw | Perpetual | Spending-rule discipline and illiquidity budget |
| Life insurer | Spread over the guaranteed crediting rate | Low, regulator-constrained | Predictable, from claims | Long, and duration-matched | Solvency capital and asset-liability matching |
| Bank | Net interest margin on the balance sheet | Low | High, deposits are callable at par | Short to medium | Liquidity coverage and interest-rate gap |
Two readings of that table are worth carrying. The first is that horizon and liquidity are different constraints and are constantly conflated. An endowment has a perpetual horizon and a real, recurring liquidity need of four to five percent a year, which is why an unconstrained frontier's answer of forty percent in illiquid alternatives can be wrong even for the investor with the longest horizon in the table. The second is that for a pension and an insurer the objective is not a return at all, it is a liability, and the right optimisation is over surplus rather than over assets. AA1.05 does that arithmetic; this node's frontier is an asset-only frontier and says so.
Which investment vehicle should hold the portfolio?
The wrapper is itself a decision. One asset allocation can be implemented through five vehicles that differ enough to change the after-tax answer.
| Vehicle | Access | Typical fee load | Tax treatment | Liquidity | Transparency |
|---|---|---|---|---|---|
| Open-end mutual fund | Retail, low minimum | Highest, and often bundled with distribution | Fund-level events pass through; in the US, realised gains distribute to remaining holders | Daily at net asset value | Periodic holdings disclosure |
| Exchange-traded fund | Retail, one share | Lowest for index exposure | In the US, in-kind creation and redemption suppresses capital-gain distributions; in India the mechanism exists but the benefit is smaller | Intraday, at a price that can differ from net asset value | Usually daily holdings |
| Separately managed account | High minimum, typically institutional or high-net-worth | Negotiated, unbundled | Owned directly, so losses can be harvested at the security level and the cost basis is the investor's own | By agreement, subject to the underlying market | Complete, position by position |
| Closed-end fund | Exchange-listed, fixed share count | Moderate to high | As for the fund type | Intraday, at a market price that can trade at a persistent discount or premium to net asset value | Periodic |
| Collective investment trust or pooled vehicle | Institutional or plan-level only | Low, negotiated | Depends on the plan wrapper around it | By plan rules, often less than daily | By agreement |
The one line to take from the table is that the separately managed account and the exchange-traded fund solve different problems. The account buys tax control, which is worth most to a taxable investor with concentrated low-basis holdings, and PW1.01 prices that. The fund buys cost and simplicity, which is worth most where the exposure is a commodity. Choosing between them on fee alone will be wrong for exactly the client who most needs advice. AA1.08 owns the exchange-traded fund's internal mechanics, including creation and redemption and the sources of tracking difference.