Tianli International Holdings

Tianli is cheap only if you count profit. Count the cash that is left after land, related-party settlements and deals, and the HKD0.73 price already matches what the business has delivered for three years. We would need to see that cash grow and reach holders before calling it a bargain.

Since 2021 Tianli's network is larger than its consolidated business. Thirty compulsory-education schools sit outside the accounts, 23 schools are only managed for a fee, and the group's revenue now comes from high-school education services, food and supplies sold to students, canteens and management fees.

The economic engine changed in 2021. Before that year Tianli owned and consolidated a chain of K-12 schools and earned tuition and boarding fees. The 2021 Implementation Rules banned control of compulsory-education schools by contract and banned their transactions with related parties, so Tianli deconsolidated 30 primary and middle schools on 31 August 2021 and wrote off RMB1,085m of the campuses they occupy. ^12,2

What remains is a service model. High-school sections were moved into new for-profit companies that the group consolidates, and the group sells comprehensive education, canteen and product services across a network that also includes managed schools. In simple terms: revenue = network students × recognised service revenue per student + product sales + canteen + management fees, and profit depends on how much of that network is consolidated and fee-paying.

The network grew 72% from autumn 2022 to autumn 2025 and then shrank for the first time since the 2021 policy reset. The fall came from a large graduating cohort and four managed schools leaving, so recovery depends on retaining students rather than on intake, which still rose.

The arithmetic of the self-operated schools is simple. They started autumn 2025 with 84,625 students and took in 28,188 new ones, yet ended with 75,866. About 36,900 students therefore left through graduation, transfer or withdrawal, more than the whole new intake. Management says it deliberately slowed intake after 2023 to protect quality, and that fewer students will graduate in 2027. ⁴

The managed-school decline matters less for profit than its size suggests, because those students pay tuition to someone else and Tianli earns only a fee. The self-operated decline matters more: those schools generate the education, canteen and product revenue. A 10% smaller self-operated base, held for a full year, is the main reason we cut FY2027 profit below the FY2026 estimate in every scenario.

Education revenue grew far more slowly than the network, so revenue per network student fell. The gap was filled by food and supplies sold at thin margins, which adds revenue and working capital but little profit.

The most plausible explanation for the falling yield is mix. The number of managed schools rose from 10 in FY2024 to 23 by February 2026, and their students count in the network while their tuition does not reach Tianli. Lower fees at newer, developing schools are the alternative explanation; the filings do not let us separate the two. Either way, the network is no longer a reliable proxy for earning power.

Deconsolidation removed the compulsory schools' profits but not their links to the group. They still use group campuses, owe and are owed large sums, and borrow with the group's guarantee. For minority holders, the legal perimeter matters more than the network.

The national classification rules that would settle the status of private schools have still not been issued, five years after the Implementation Rules took effect. The company says the uncertainty remains substantial. For Tianli the practical consequence is that the 30 schools stay in limbo: they cannot pay the group, yet their balances with it are being settled in cash and the group still lends to them. ²

Hong Kong-listed China K-12 operators all trade at low multiples, and Tianli sits in the cheap cluster on profit. On operating cash flow it is priced above the peer median, which says the market already discounts the gap between its profit and its cash.

The closest comparable is Virscend, another Sichuan K-12 group reshaped by the 2021 rules, which trades at a higher P/E but a lower multiple of cash flow. Maple Leaf and YuHua sit even lower on both measures. The sector lesson is that investors price these companies on cash they can see reaching shareholders, and Tianli's next re-rating has to come from that, not from closing a P/E gap.

Gross margin was remarkably stable through four years of growth, helped by government grants. The profit warning shows the second half of FY2026 losing more than half of last year's profit, because new costs arrived just as students fell.

Some of the new costs are choices rather than structural pressures: AI products, the Hong Kong school and consultants can be scaled back. The higher expected-credit-loss charge is different, because it points to receivables that may not be collected, and the largest receivable on the balance sheet is the RMB873m owed by the deconsolidated schools. The company has not said which receivables drove the charge.

Operating cash flow is strong, but land, campus capex and interest absorb about half of it. In FY2025 the cash that was left went, more than twice over, to settle balances with the deconsolidated schools; banks and the cash balance funded dividends.

The balance sheet has not grown more risky in total; it has swapped an interest-free claim from related schools for interest-bearing bank loans. That swap is close to finished, which is the main route by which cash could start reaching holders again.

Tianli paid rising cash dividends until FY2025 and its chairman bought shares in early 2026. Since then the board has stopped the interim dividend, spent RMB452m on land and committed RMB245m to an acquisition at four times its own multiple.

Two other flows belong in the record. Nanyuan Construction, owned by the chairman's holding company, builds campuses at cost plus 9–11% (RMB60.7m in FY2025, RMB50.1m in the first half of FY2026). And the chairman added about 30 million shares between September 2025 and February 2026, the last at HKD2.73–2.79, now 74% under water. ^1,2,10

Insider buying followed by a dividend stop can precede a cheap accumulation or a restructuring. Here the timing argues the other way: the chairman and the company bought at about four times today's price, before the stop. The observable explanation is that the cash was needed for land and the deal; we infer no motive beyond that.

The dominant pattern is a related-party cash drain on a shrinking base: the profit is real, but the cash flows first to claims, land and deals outside minority control. The discount is mostly deserved today; it becomes excessive only once the remaining RMB0.6bn claim is settled without new ones.

What the price already assumes

At HKD0.73 the shares are worth RMB1.31bn. Add the claims that rank ahead of holders in our base case (the RMB602m related-party claim, the net cost of the Tianlai deal and an expected loss on guarantees) and capitalise at a 15% cost of equity with no growth: the market is paying for about RMB305m a year of sustainable cash after reinvestment. That is slightly more than the RMB287m Tianli produced on average in FY2023–FY2025.

So the market is not pricing collapse. It is pricing three fears: that students and profit keep falling from here, that cash keeps leaking to related parties and deals, and that a controller who sits on both sides of those flows decides what is left. The issues below test each fear.

Why is it cheap: temporary or permanent?

The cause that decides value is the related-party drain, and on the evidence it is temporary but not yet proven so. Most of it is a legacy debt being repaid: the net claim has fallen 73.2% since August 2023 to RMB0.60bn. But new advances of RMB184m in the first half of FY2026 and new guarantees show it can restart. The regulatory cause is the 2021 rules that took the compulsory-stage schools out of the group; that is temporary by default, and the company can wait, but our values give no credit for a reversal. The demand cause is open: students served fell 14.5% while new intake rose 4.6%.

Cheap or value trap?

Named pattern. The Related-Party Cash Tunnel, in a legacy-claim form: profit is reported by the group, but cash moves to related schools through debt settlement, advances and guarantees. Related-party cash flows above 10% of revenue trigger a full-audit flag in our framework; Tianli's were 21% in FY2025 and 15% in the first half of FY2026. A second pattern, the licence ownership gap, applies to the 30 schools whose licences and profits sit outside the group while its campuses serve them.

Deserved versus excessive. The deserved part of the discount is the RMB0.73bn of claims, the shrinking student base and cash conversion of about half. We cannot show an excessive part today, because our base value is below the price. The discount becomes excessive if the claim is settled without new advances and the census stabilises; then about RMB300m a year could be paid out, a 23% yield on today's market value. What could block value permanently is new related-party funding or further acquisitions that absorb that cash.

We value the cash that could reach holders, after reinvestment and interest, and deduct the claims that come first. On that basis the shares are about fairly priced, with a wide and asymmetric range around the base case.

Risk-reward at HKD0.73: no reward to the base case. The base value is 8.2% below the price and the bear case 79.5% below, so neither entry gate passes, and the probability-weighted value is about equal to the price. The full panel is on the decision page.

Method. We capitalise sustainable cash after reinvestment, interest and leases at a cost of equity of 13–18%. From that value we deduct the net related-party claim of RMB602m (interest-free and payable within one to three years, so taken at face value), the Tianlai cost net of the value of Tianli's share of its profit target, and an expected loss on the RMB254m of guarantees. We divide by 2,088.5m shares; the 67.8m share options are out of the money (exercise prices HKD2.38–4.72) and are ignored. A 5% move in the renminbi changes the HKD value by about 5%.

We reject reported P/E as the primary method because it ignores both the claims and the 50% cash conversion, and we reject peer multiples because the peers' perimeters and policy positions differ too much. EV/EBITDA is shown only as a cross-check: about 3.7 times, distorted by grants and the related-party claim.

The largest risk is not solvency but leakage: if the deconsolidated schools or new deals absorb cash again while students fall, the bear case of about HKD0.15 becomes the central case. The FY2026 annual report at the end of November is the first test.

• Upgrade if: the FY2026 annual report shows the net related-party claim below RMB400m with net repayments to the group, second-half land spending below RMB100m (an FY2026 total below about RMB552m, as RMB452m was spent in the first half) and a final cash dividend; and FY2027 is then on course for cash after reinvestment of at least RMB350m with land below RMB100m for the year. A stable autumn 2027 census would then support Deep-Value Recovery.

• Downgrade to Value Trap if: advances to the deconsolidated schools exceed repayments again, a new guarantee is given, or another acquisition above RMB200m is announced while no cash dividend is paid.

• Kill trigger: a regulatory action against the structured contracts, a withdrawal of bank facilities or a covenant breach, or a qualified audit opinion.

• Price and position-size discipline: no position at HKD0.73. Below about HKD0.47, 30% under the base value, a small starter position would be justified on value alone; full size only after the upgrade conditions. Daily turnover of about 3 million shares (about HKD2m) limits position size in any case.