Why Mega-Cap Concentration Is a Portfolio Decision, Not a Market Condition
Index concentration is usually discussed as weather. It is a position — one you either sized deliberately or inherited by default, and only one of those is examined.
Index concentration is almost always discussed as if it were weather — a condition of the market to be commented on, worried about, or waited out. That framing contains a quiet error. The share of your equity portfolio sitting in the largest handful of companies is not a description of the environment; it is a number on your own position sheet, and you either chose it or accepted it by default.
Both are decisions. Only one has been examined. This piece is about how an institutional risk desk turns the second kind into the first.
The Default Is a Position
A capitalization-weighted index does not target diversification. It targets ownership proportionality: every holder's weights sum to the market, and the weights are an output of past price behavior rather than a forecast of future risk. That mechanical property has a consequence most owners never state out loud. When a group of companies compounds faster than the rest of the market, the index does not trim them. It hands you more of them, continuously, without asking, and it will keep doing so for as long as the trend runs.
So the portfolio you hold at any moment reflects an accumulated series of non-decisions. Nothing about that is inherently wrong — inertia is often the correct policy, and the historical record of deviating from cap weighting to "manage" concentration is not a happy one. The problem is that an unexamined position cannot be sized. You cannot know whether you are comfortable with a risk you have never measured, and you certainly cannot know how it interacts with the rest of what you own.
Name Count Is Not Diversification
The most common defense of a concentrated index is that it still holds hundreds of companies. This confuses two different things. Diversification is not a count of legal entities; it is a count of independent sources of return. If a large share of portfolio value sits in businesses that share a customer base, a capital cycle, a sensitivity to long-duration discount rates, and a regulatory perimeter, then the number of tickers overstates the number of bets.
There is a second, more practical version of the same error, and it is the one that does real damage in multi-manager portfolios. Concentration compounds across sleeves. A domestic core allocation, a growth fund, a global equity mandate, a quality-factor product and a target-date vehicle can each look reasonably diversified in isolation while holding substantially the same top names. The owner's true exposure is the look-through sum, not the exposure disclosed by any single line item. Institutions run this aggregation as a matter of routine; individual and advisory portfolios frequently do not, and are therefore carrying a concentration they have not seen.
Volatility Is Not Risk — and Concentration Exploits the Difference
Volatility is the observed dispersion of returns over a sample. Risk is the probability of permanent impairment or of failing to meet the obligation the money exists to fund. Most of the time these travel together, which is why the industry uses one as a proxy for the other. Concentration is one of the situations where they can separate, and the separation runs in the uncomfortable direction.
Consider the mechanics. A group of large, profitable, cash-generative companies that grinds steadily higher lowers measured index volatility while it does so — a smoothly rising heavyweight suppresses the standard deviation of the whole. At the same time, it raises the share of portfolio outcomes that depend on a small number of business models. Backward-looking risk statistics will therefore tend to understate concentration risk exactly when concentration is greatest, because the very price behavior that built the weight is the behavior that flatters the volatility estimate. Any risk framework that trusts trailing volatility as its primary input has a structural blind spot here, and it is not a subtle one.
What Correlation Does Under Stress
The second reason concentration is a risk question rather than a volatility question is that correlation is not a constant. In calm markets, stock-level dispersion is wide and idiosyncratic drivers dominate; a portfolio genuinely behaves like a collection of separate bets. In drawdowns, dispersion tends to collapse and cross-sectional correlation rises toward one as investors sell exposure rather than companies. Diversification, in other words, has historically been at its weakest at the moment it is most needed. A full-sample correlation matrix averages the good regime and the bad regime together and produces a number that describes neither.
Market history offers two illustrations worth holding in mind, both qualitative and both well established. The unwind of the late-1990s technology leadership showed that a concentrated index can deliver a broad-market drawdown even when the median stock fares considerably better — the arithmetic of weight cuts in both directions. The 2022 repricing showed something subtler and arguably more important for portfolio construction: when the largest index constituents are long-duration assets whose valuations are sensitive to discount rates, they share a driver with the bond allocation that was supposed to hedge them. Equity concentration in long-duration businesses quietly converts a two-asset portfolio into something closer to a single macro bet on rates. That is a correlation problem, not a stock-selection problem, and it is invisible if you only inspect the equity sleeve.
The Strongest Case Against This Framework
The serious counterargument is not that concentration is harmless. It is that concentration is earned. Economic profit in a modern economy is itself highly concentrated; a small number of firms with scale advantages, network effects and net-cash balance sheets generate a disproportionate share of aggregate corporate cash flow. On that view, a cap-weighted index is simply reporting a fact about the economy, and the largest constituents carry less idiosyncratic failure risk than the average smaller company, not more. It follows that cap weighting is the neutral, market-clearing portfolio, and that any deviation — an equal-weight tilt, a cap on single-name exposure — is the active decision requiring justification.
Most of that is correct, and this framework does not dispute it. Note what it does and does not establish. It establishes that concentration can be fundamentally justified; it does not establish that any particular owner should hold it in the size the index happens to deliver. The market-portfolio argument is cleanest for an investor with an infinite horizon, no liabilities, and no other assets. Real owners have spending dates, and they frequently hold correlated exposures elsewhere — employer equity, a private business in the same value chain, real estate in a metro whose employment base is the same handful of firms, a bond portfolio sensitive to the same discount rate. For those owners, the index's weights are an input to the decision, not the answer. And the honest version of the argument cuts both ways: underweighting the leaders has been expensive during leadership runs, which is precisely why the underweight also deserves to be a sized, deliberate position rather than a reflex.
How Institutions Size the Bet
Desks that handle this well rarely ask whether concentration is "too high." The question is not answerable in the abstract. They ask instead how much of total portfolio risk — not total portfolio value — is attributable to a small set of positions, and whether that allocation of the risk budget was intended. A useful diagnostic is to decompose active risk versus the benchmark and see what fraction comes from the own/underown decision on a handful of names. When a single relative-weight decision accounts for a dominant share of tracking error, the portfolio has effectively become a bet on that decision regardless of what the holdings page suggests it is doing.
The second discipline is procedural rather than analytical: the policy has to be written before it is needed. Cap weighting never generates a trim signal, so in the absence of stated rebalancing bands, drift is unbounded and the portfolio's risk profile is set by price action rather than by the owner. Bands and tolerances chosen in advance are not a forecast; they are a commitment device that converts a running series of non-decisions back into a policy. Equal-weight and completion portfolios are best used first as measurement tools — how different does the world look without the weight? — and only second, if at all, as allocations.
How to apply this framework
- Aggregate look-through weights across every sleeve. Sum your exposure to the top names across all funds, mandates, employer equity and private holdings. If the combined figure surprises you, the concentration was inherited rather than chosen — and that gap is the finding, not the number itself.
- Measure risk contribution, not just weight. Weight tells you what you own; contribution to variance tells you what drives outcomes. When a small number of positions dominate the risk decomposition, the portfolio is more concentrated than the holdings page implies.
- Test correlations in drawdown subsamples, not the full sample. A diversification assumption that only holds in calm regimes is not a diversification assumption. Ask specifically whether your hedging assets share a driver — duration, credit, a single macro variable — with your largest equity exposures.
- Write the rebalancing bands before you need them. Cap weighting will never ask you to trim. A stated tolerance, set in advance and reviewed on a calendar rather than on news, is what turns concentration from a drift into a deliberately held position — whichever side of it you choose to be on.