How Korea's National Pension Fund Builds an 'Efficient Portfolio'

South Korea's National Pension Service (NPS) — the country's public pension fund and one of the world's three largest — is often called a textbook of “efficient diversification.” Does that hold up in numbers? We ran the NPS's top 100 equity holdings (year-end 2024) through DIVA Quantizer to examine Markowitz's Modern Portfolio Theory in practice.

Korea's National Pension Fund manages well over ₩1,000 trillion (≈ US$700 billion). A pool that size could hardly be poured into a handful of “good-looking” stocks. Its philosophy isn't about picking winners; its core is Strategic Asset Allocation (SAA) — first deciding the split across equities, bonds and alternatives, then diversifying risk within each.

The root of this is Harry Markowitz's Modern Portfolio Theory (MPT, 1952). Its insight is simple: don't ask “which stock is good,” ask “how do the holdings move together.” When one asset wobbles while another stays put or moves the other way, holding both shrinks the swings of the whole portfolio — cutting risk without giving up return. About as close to a “free lunch” as investing gets.

A pension fund's broad diversification is MPT put into practice. So is the NPS's equity portfolio really “efficiently diversified”? Let's check with numbers, not words.

The analysis uses the holdings the NPS disclosed for year-end 2024. From its Korean and US-listed holdings that DIVA supports, we took the top 100 by market value — about 47.5% (≈ ₩267 trillion) of the NPS's total equity value. Stocks that listed after 2022 and would shorten the window (LG Energy Solution, HD Hyundai Heavy Industries, Krafton, etc.) were excluded, leaving names with a full six-year (72-month) price history. Each was weighted by market value into one portfolio and run through DIVA's Markowitz analysis as-is.

Diversification was genuinely at work

The most striking figure is the diversification benefit of 39.3%. Held individually these 100 stocks average 31.9% annual volatility; combined by weight, that drops to 19.3%. Because Korean semiconductors, US big tech, staples, financials and healthcare don't all move the same way on the same day, putting them together shaves nearly 40% off the risk — the textbook diversification effect, intact, in a real large-scale portfolio.

The Sharpe ratio — return per unit of risk — came in at 1.16, an “excellent” level (the past six years were broadly a bull market). On the frontier chart the NPS's point (red dot) sits below the curve, but that curve is an “optimistic” estimate, its top lifted by a few surging names like NVDA — because Markowitz projects past average returns straight into the future. Against an estimation-error-corrected benchmark, this much diversification and Sharpe is comfortably efficient, which is why DIVA labels it “The Impregnable Fortress” — broad diversification and low volatility forming a line that holds firm against any risk.

Source: National Pension Service Fund Management disclosure, “Equity Holdings by Issue (year-end 2024).” Prices: yfinance month-end adjusted close (2019–2024, 72 months); risk-free rate: year-end-2024 US 13-week T-bill (4.21% annual).

Giant pension funds aren't the only ones who benefit from diversification — the same principle applies to an individual holding a handful of stocks. Three practical takeaways from the analysis above.

① How to read your position on the frontier

The efficient frontier connects “the maximum return possible at a given risk.” A dot sitting far below it can signal room to adjust weights. But the curve itself is an “optimistic” estimate, inflated by past high-flyers — so DIVA grades archetype efficiency against an estimation-error-corrected benchmark. The lesson: a periodic “efficiency check” for individuals should weigh not just distance from the curve, but corrected risk-adjusted efficiency (Sharpe) too.

② Efficiency isn't everything

That the NPS's dot sits below the curve doesn't mean “its strategy is wrong.” Markowitz looks at one yardstick: the efficiency of past return relative to volatility. But the real NPS blends in many judgments and constraints — macro outlook, currency, liquidity, policy limits, and maturity-matching against long-term pension liabilities. Not being optimal on that single axis is closer to a sign that the other axes are operating too. And this analysis fed in only part of the equity book (top 100), which adds its own error. “Below the frontier,” then, is less proof of inefficiency than a picture of real-world management that efficiency alone can't explain.

③ Diversification is about ‘correlation,’ not the ‘number of holdings’

The NPS's risk fell not because it holds many stocks, but because it mixed assets that move differently (Korean + US equities; chips + staples + financials + healthcare). Gather 100 stocks that move alike and you get little diversification. So how do you check whether your portfolio is secretly betting on ‘one and the same story’? DIVA Quantizer's ‘Risk Balance Check’ (HRP model) is built for exactly that — it auto-clusters holdings that move alike, revealing whether risk piles up on one side even when things look well diversified.

Even one of the world's largest pension funds rests on a sentence Markowitz offered ~70 years ago — “a security's risk has no meaning on its own; what matters is its contribution to the whole portfolio.” Checking those relationships in your own portfolio, in numbers, now takes a few clicks.

This article was written with the help of AI.