A Lower Yield Can Add More to a Portfolio

With bond-equity correlation staying elevated,¹ investors may look for additional diversifiers. Private markets are natural candidates, but whether a given private investment truly diversifies, and adds enough value to be worth the cost, depends on both the asset and the portfolio already in place.

A total-portfolio approach starts from a plain principle: Every investment has to earn its place by what it adds to the whole portfolio, rather than by how it performs on its own. A companion post showed how to test that, by comparing an asset's expected return with the hurdle rate — the return a new asset has to earn to justify the risk it adds to this particular portfolio.² But clearing that bar is only the first step.

Consider a hypothetical Canadian defined-contribution (DC) pension plan of about CAD 12 billion, invested for members who are decades from retirement. Its portfolio is entirely public: 55% in public equities, 40% in fixed income and 5% in listed real-estate investment trusts (REITs). The board has agreed to look at private real estate, which raises three questions: Which market should the plan buy, how large should the position be and what can the manager sell to fund it?

A total-portfolio approach judges each candidate by its contribution to the whole, captured by the hurdle rate. We screened five markets: Canada, the U.S., the U.K., Australia and Japan. We represented each market with an MSCI Private Real Estate Index³ and mapped each onto the MSCI Multi-Asset Class (MAC) Factor Model in BarraOne® to place public and private assets on one basis. Mapping to a factor model is crucial for correctly capturing the correlation between public and private markets.⁴

Foreign property introduces foreign-currency risk, and that risk can easily swamp the property decision it is attached to. In a total-portfolio approach, currency is typically managed at the total-fund level, rather than asset by asset.⁵ To keep the screening focused on the property decision and to avoid introducing currency exposure the plan does not want, we hedge the foreign candidates back to Canadian dollars.⁶

We calculated the hurdle rates for all assets and compared those to their respective expected returns.⁷ All five markets cleared, but for private assets the factor-model-implied hurdle is only part of the test. The plan may also demand a premium for capital lockup and the governance cost of running a private allocation. For our example, even at 200 basis points above the hurdle, all five markets still clear.

The most additive market isn't the highest-yielding

The most additive market is not the one with the highest expected return, in this simulation. The U.K. has the highest stand-alone return, at 7.0%, yet ranks only third, because its high correlation and volatility push up the hurdle. Japan is the reverse. It diversifies well but yields too little to clear by much, which shows that low correlation alone does not earn a place: The return still has to cover the hurdle. Australia wins on both counts. Its property is less correlated with what the plan holds, and it still offers a competitive return. The hurdle rate turns diversification benefit into the same units as expected return: The lower the hurdle, the more genuine the diversification. The gap between the two is what sets the markets apart.

The efficient frontier tells the same story. Adding 5% of each market and optimizing the rest, Australian property produced the highest frontier and Japan the lowest across the whole range.

Efficient frontiers were constructed via mean-variance optimization. For portfolio assets, we used their portfolio-implied returns assuming their initial weights and initial 0.5 Sharpe ratio; for the candidate assets, we used the expected returns described above. The allocation to private real estate was fixed at 5%. The asset covariance matrix was constructed using the extra-long-horizon version of the MSCI MAC Factor Model, as of July 1, 2026.

To size the position, we held total risk at the level the plan already runs and read off the extra return.

Efficient frontiers were constructed via mean-variance optimization. For portfolio assets, we used their portfolio-implied returns assuming their initial weights; for the candidate asset, we used the expected returns described above. The allocation to private real estate was fixed at 5%, 7.5% or 10%. The asset covariance matrix was constructed using the extra-long-horizon version of the MSCI MAC Factor Model, as of July 1, 2026.

With risk unchanged, adding Australian real estate lifted expected return by about 27 basis points at a 5% allocation, 39 at 7.5% and 52 at 10%. On risk-return grounds alone the largest allocation wins in this simulation, but the pension plan also weighs liquidity: Members may transfer out, and the plan needs to rebalance without forced sales in a market with no daily pricing. Governance capacity matters too: A small team overseeing its first private-markets allocation may prefer a position it can monitor and explain to the board without strain.

We looked at how the optimizer funded each allocation level. The pattern is consistent: Most of the capital comes from fixed income, with a smaller trim to equities and a marginal cut to listed REITs. The private-real-estate allocation acts as a diversifier, so the optimizer can afford to pull from the lowest-returning sleeve. A more correlated addition would have forced a deeper cut to equities instead.

The approach is not specific to an asset class or an initial portfolio composition. Any new allocation, public or private, can be structured through the same sequence: Calculate the hurdle rate, test the fit and decide how to fund.