MEL Fund

Limits and next steps

Read the v1 conclusions with these limits in mind.

Limits of the v1 model

  1. Narrow liabilities. Only personal balances, payments owed to leavers and the shared part of pensions in payment count as liabilities — not future education and medical benefits — so solvency reads high. Defaults are detected mainly by unpaid benefits.
  2. No medical cost growth. Amounts are real values and neither medical costs nor incomes grow. If medical costs grew 1–2% a year faster than incomes, the medical pool could slide into deficit late in the period: in years 31–49 its spending (720 yuan per contributor) already exceeds its income (655 yuan).
  3. Joining and leaving are assumptions. The 1.5% yearly join rate, 3% exit rate and the rest have no local data behind them. The 13% voluntary ceiling depends directly on them.
  4. Fixed shock timing. All stress shocks come after the switch to mandatory membership. A shock in the switch year itself has not been tested.
  5. Seed split 30/40/30. The gap is in the education pool; putting all seed money there would lower the amount needed.
  6. Extrapolated population. Residents fall from about 195,000 to about 157,000 because low fertility and youth out-migration continue.
  7. Few runs. 50 runs per scenario and a seed resolution of about 4 million yuan. Firm conclusions need 200 runs or more.
  8. The OCR data has not been checked by hand, only by its sums.

Next steps (v2)

  1. A stress test with medical costs growing 1–2% a year faster than incomes — the top priority.
  2. Future education and medical benefits counted as liabilities, with solvency reported for a closed and an open population.
  3. A shock in the switch year, and all seed money in the education pool.
  4. Exit as a threshold function of cost: almost no exits while the burden is low, mass exits above a threshold (estimated from Xianghubao data).
  5. Household membership as a comparison scenario.
  6. The same calibration for the control county — many old people and few children, the opposite of the study county.
  7. 200 runs or more per scenario.

The research report will have a separate chapter that asks only whether this kind of fund can exist legally:

  • Could it be treated as illegal fundraising?
  • Does it need an insurance licence?
  • Which legal form could carry it: cooperative, trust, charity, mutual society?

That chapter has not started. Until it is done, MEL Fund is research, not a fund anyone can join.