Are the stocks Korea's pension funds buy
any different?
"Institutions" is not one thing. Korea publishes daily, per-stock net buying for pension funds, investment trusts, insurers, private funds, banks and securities firms' own books — separately. After testing foreigners and retail, I split institutions seven ways and counted six years. If the National Pension Service is buying, surely that means something.
01The data and the rules
KOSPI is Korea's main board, KOSDAQ the secondary one; the session runs 09:00–15:30 Korea Standard Time (UTC+9). After every close the Korea Exchange publishes net buying by investor category for every stock. Beyond retail and foreigners, the institutional side is broken into pension funds (dominated by the National Pension Service, one of the world's largest pension funds), investment trusts (mutual funds), insurers, private funds, banks, securities firms (proprietary books) and other corporations. I measured all of them with the same yardstick, foreigners and retail alongside for comparison. Since the figure is only known after the close, every trade is at the next open.
- Net buy — any positive amount that day
- Heavy net buy — net buying of at least 3% of that day's volume
- 20-day accumulation — net buying over the past 20 days of at least 2% of 20-day volume (to catch slow accumulators like pension funds)
Buy the next open; sell at the open 1 week, 1 month or 3 months later; 0.22% deducted on every buy. The baseline is the same 749 stocks on any random day. The median is the middle result when every trade is lined up from worst to best.
02How often each one buys
| Investor type | Net buyer | Traded at all |
|---|---|---|
| Pension funds | 25.1% | 52.8% |
| Investment trusts | 28.4% | 61.6% |
| Insurers | 19.5% | 47.0% |
| Private funds | 27.3% | 59.9% |
| Banks | 5.6% | 11.9% |
| Securities firms (own books) | 39.5% | 75.4% |
| Foreigners | 47.6% | 98.2% |
| Retail | 50.5% | 98.5% |
Pension funds do not touch half of all stock-days. Banks are barely present and are dropped from the tables below.
03Which days each one buys on
| Investor type | Net-buy days up | Median | Heavy-buy days up | Median |
|---|---|---|---|---|
| Pension funds | 57.0% | +0.42% | 65.1% | +0.68% |
| Investment trusts | 59.4% | +0.57% | 70.6% | +1.04% |
| Insurers | 57.1% | +0.44% | 69.1% | +0.79% |
| Private funds | 55.9% | +0.38% | 66.5% | +0.81% |
| Securities firms | 51.5% | +0.15% | 60.3% | +0.43% |
| Foreigners | 59.2% | +0.51% | 66.8% | +0.86% |
| Retail | 27.6% | −0.86% | 11.7% | −1.52% |
Everyone except retail buys on up days. When investment trusts buy heavily, seven days in ten are up days, median +1.0%. Pension funds: 65%. Retail alone is on the other side, buying the down days.
This is not a verdict on who is right. Institutions and foreigners buy what is rising (trend following); retail supplies the shares. The only question left is: does the stock they bought on an up day keep rising?
04Following them
| Investor type | Count | 1-month win rate | 1-month median | 3-month win rate | 3-month median |
|---|---|---|---|---|---|
| Pension funds | 5,820 | 44.6% | −1.23% | 42.0% | −2.70% |
| Investment trusts | 2,941 | 44.2% | −1.29% | 43.2% | −2.36% |
| Insurers | 528 | 44.5% | −0.69% | 44.3% | −1.59% |
| Private funds | 1,817 | 45.7% | −0.78% | 45.1% | −2.28% |
| Securities firms | 6,487 | 44.9% | −0.95% | 46.6% | −1.36% |
| Foreigners | 18,616 | 46.9% | −0.77% | 46.8% | −1.31% |
| Retail | 25,122 | 46.5% | −0.66% | 45.8% | −1.62% |
Not one of the nine beats the baseline. Whoever bought heavily, buying the same stock the next morning did worse than random over three months. The worst of all is pension funds: 42.0% winners, median −2.7%, against 48.4% and −0.45% for a random purchase.
The 20-day accumulation signal, built to catch slow buyers, is worse still. Stocks that pension funds accumulated to more than 2% of 20-day volume won 37.1% of the time three months later, median −3.7% — 11 points below the baseline. Investment trusts 42.1%, private funds 43.6%.
Widen to "any net buy" and everything returns to within a point of the baseline, just as in the retail article: the stronger the flow, the worse; the weaker, the less it says.
05Why pension funds are the worst
Pension funds buy stocks that have risen, in size, over several days. Section 03 showed 65% of their heavy-buy days were up days; by the 20-day measure the stock has already climbed for most of those 20 days. Whoever buys the next morning starts from the price the pension fund built.
A pension fund tracks an index, rebalances to target weights, and spreads purchases over months. Its buying says less "this stock is good" than "this stock is underweight". So pension-fund net buying describes the fund's past purchases, not the stock's future — and the data says the price mean-reverted once the buying stopped.
Put the foreign, retail and institutional articles together and they say one sentence: a day of heavy flow from anyone is a day the stock moved a lot, and stocks that moved a lot did worse over the next three months. Who did the buying was secondary.
06Caveats
749 stocks above 100 billion won with six years of history. Pension funds are large-cap investors so the sample is adequate, but small caps may differ.
"Heavy" signals run from hundreds to a few thousand cases — insurers 528, banks 40 (dropped). Seven types pointing the same way is hard to call chance.
Net buying is in shares. Costs are the 0.22% buy-side deduction only. Delisted stocks are not included.
07What I take from it
- Everyone except retail buys on up days: 71% of investment trusts' heavy-buy days and 65% of pension funds' were up days
- Whoever bought heavily, following them the next morning lost to the baseline over three months. All nine types
- Pension funds were the worst: 42.0% on heavy buys, 37.1% on 20-day accumulation (baseline 48.4%)
- "Any net buy" carries no information — within a point of baseline
- Across three articles on flow, one conclusion: a stock that moved a lot did worse afterwards, whoever bought it
"The pension fund is buying" turned out to mean "this stock has already gone up". Whoever followed was not buying alongside the pension fund; they were buying the price the pension fund made, and that price gave ground for three months.
댓글 쓰기