Diversification carries risks Canadian advisors rarely name to clients

A client walks into your office with holdings in four Canadian banks, two energy producers, and a REIT. They believe they are diversified. The correlation between those positions during a rate shock or a commodity downturn approaches 0.95, meaning they move together, not apart. What looks like diversification is sector concentration wearing a costume.

The standard advice, spread capital across asset classes, industries, and geographies, rests on a simple mechanism: low or negative correlation reduces volatility. When equities fall, bonds rise. When Canadian financials lag, U.S. tech compensates. The math works until it doesn't, and the failure modes get named far less often than the benefits.

The correlation convergence problem

Modern Portfolio Theory assumes that correlations are stable. They are not. During the 2008 crisis and again in March 2020, correlations between equities, corporate bonds, and real assets all spiked toward 1.0 within days. Diversification did not reduce risk when it was needed most. It evaporated. The portfolio that was designed to dampen volatility across twelve independent sleeves became twelve versions of the same trade, all moving down together. This is not a flaw in the theory. It is a feature of how markets behave under systemic stress.

The S&P/TSX Composite Index sits at roughly 31% in Financials and 18% in Energy. A client who diversifies by adding more TSX names is not solving for correlation risk. They are compounding it. Real geographic diversification requires holdings that do not share the same regulatory environment, central bank policy, or currency risk. A position in U.S. equities is not just an equity allocation. It is a deliberate hedge against a declining Canadian dollar and exposure to sectors, technology, healthcare, consumer discretionary, that barely exist at scale in Canada.

The cost of over-diversification

There is a point past which adding assets provides no further risk reduction but adds cost and complexity. This is called diworsification. A portfolio holding 80 positions across 12 asset classes and 9 geographies may look sophisticated. What it often represents is a 2.1% blended MER applied to a risk profile that could have been achieved with 22 holdings.

The math is not symmetrical. The reduction in standard deviation from moving your portfolio from 10 holdings to 30 is measurable. The reduction from 30 to 80 is negligible. Meanwhile, the tracking error introduced by niche allocations, frontier market debt, bitcoin futures, liquid alts, can produce volatility spikes that exceed what diversification was meant to suppress. Advisors who add sleeves because they are available, not because they solve a correlation problem the client actually has, are mistaking activity for progress.

Rebalancing is not optional

Diversification does not maintain itself. A 60/40 portfolio that is not rebalanced drifts toward 70/30 or 50/50 depending on which sleeve outperforms. The drift changes the risk profile. A client who signed up for moderate volatility in 2021 is now holding an aggressive allocation in 2026 without ever making an active decision. Rebalancing forces the sale of outperformers and the purchase of underweights, which feels wrong but is mechanically necessary.

The psychological cost is real. Clients resist selling winners. Advisors who frame rebalancing as a risk-management requirement rather than a trade are more likely to retain compliance. Systematic rebalancing schedules, annual, semi-annual, or tolerance-band-triggered, remove discretion and reduce the emotional friction of the decision.

Diversification works. It reduces unsystemic risk and smooths returns across time. But it does not eliminate systemic risk, it can fail during liquidity crises, it stops helping past a threshold of complexity, and it requires active maintenance. The clients who understand those limits make better decisions than the ones who think the work stops once the allocation is built.