Roger Urwin has been around longer than me, and I’ve been looking at and learning from his work for at least a quarter of a century. He’s still sharing his thinking (sophisticated as it is) with those who have time to think about Total Portfolio Approach (TPA)
I suspect that we will end up contracting out the investment of Pensions Mutual collective pension (CDC) and so Roger’s thinking is for someone else but if they can play back to us his thinking, I suspect that I’ll be a lot wiser and more confident of what they’re up to!
Thanks Roger – I’m making this available to my investment friends!
Thinking ahead?
There’s much to like in the 19-page WTW paper accessible via LinkedIn.
“Investment management has become extraordinarily sophisticated at analysing parts of the world while potentially becoming less capable of understanding the world as a whole.”
But the solution should not necessarily be for pension schemes to give asset managers an ever-expanding mandate.
Indeed, there is a danger that the paper’s proposed cure creates another form of overconfidence. Do asset managers possess either the democratic legitimacy or the epistemic competence to become system-level leaders? Some humility wouldn’t go amiss.
I then ask myself why “beta/alpha”
thinking still persists after all these years (and whether there’s a better way to use it without trustees feeling trapped) …
Beta/alpha persists because it’s embedded in manager mandates, flawed benchmarks (so I can agree with Mr Urwin regarding trustee failures when trying to optimise “individual parts relative to benchmarks”), and shallow reporting.
Beta/alpha continues to be a quick way to talk about “market vs skill” or “passive vs so-called active” in know-all committees or consultant-led groupthink.
For me, it’s better treated only as a reporting overlay on top of a deeper decomposition analysis.
I can still look at beta/alpha numbers if I have to, but prefer to ask: “What yield/growth/re‑rating/currency and any other assumptions are hidden inside this beta, and what specific bets make up this alpha?”
I’d favour, instead of a system-level approach presuming hyperskill, something along these more mundane lines:
Asset‑level return build‑up
For each major asset class, at the level of each security or holding, for each investment manager in the portfolio, using money-weighted (not time-weighted) returns.
eg Listed equities
• Starting yield (dividends + buybacks).
• Real earnings growth assumptions.
• Inflation pass‑throughs.
• Expected multiple change over the horizon (conservative/base/optimistic).
• Currency exposure and hedging policy.
eg Public credit (IG/HY/EM debt)
• Starting yield to worst /YTM.
• Expected defaults and recoveries.
• Spread change assumptions.
• Duration and rate path assumptions.
• Currency effects.
eg Private equity/infrastructure/other real assets (including rental properties)
• Cash yield (distributions).
• Underlying cashflow growth assumptions.
• Entry/exit multiple assumptions (and sensitivity).
• Illiquidity premium and leverage effects.
• Currency and inflation linkages.
eg Liability‑hedging assets (gilts, swaps, etc.)
• Yield and duration.
• Inflation linkages.
• Hedging costs.
Portfolio aggregation and stress testing
Then aggregate at total portfolio level, but still keeping the decomposition visible:
• Portfolio yield (cash + synthetic).
• Portfolio growth (real cashflow growth).
• Portfolio re‑rating sensitivity (how much of expected return depends on multiple expansion or spread compression).
• Portfolio currency exposure (net open FX by currency).
• Fees and illiquidity drag as explicit negative components.
Stress scenario discussions then become more transparent:
• “If equity multiples fall by 20% over 5 years, how much does expected return drop?”
• “If private market exit multiples are 15% lower than base case, what happens to the private premium?”
• “If GBP weakens by 10%, how does that affect total return?”
That gives me a genuine margin‑of‑safety analysis. I can see more clearly which assumptions are doing the heavy lifting and where some of the vulnerabilities are.
I’m afraid busy diagrams and pretty pictures are just that.
I prefer detailed portfolio arithmetic.
It’s hardly what I’d call “mathematics”, although beta/alpha probably is, if it’s not bordering on pseudoscience.