Article
A bank doesn't re-size the loan for every stress. It runs the loan it agreed to lend.
So the engine freezes Riverbend's R554.9m at financial close (principal and the sculpted repayment schedule) and re-runs everything else: interest, the DSRA, tax, the waterfall.
Minimum DSCR, debt frozen:
| Base case | 1.57x |
|---|---|
| P90 one-year | 1.45x |
| P99 one-year (86.4% of P50) | 1.34x |
| Tariff -10% | 1.40x |
| Opex +10% | 1.54x |
| Base rate +200 bps | 1.53x |
| Capex +20%, equity funded | 1.57x |
| Combined (P90, capex and opex +10%, rates +100 bps) | 1.40x |
| COD delayed six months | 0.67x |
The P99 comes from the yield report's P90, treating annual output as normally distributed: sigma = (1 - 0.925) / 1.2816 = 5.85%, and P99 = 1 - 2.3263 x sigma.
Every case but the delay holds above the 1.05x event of default, and the combined downside stays above the 1.15x lock-up.
The repayment profile is fixed at financial close, and the first repayment falls six months after COD but no later than a set backstop date. A six-month delay leaves half a year of revenue missing when those instalments fall due. That gap belongs to the DSRA, the EPC contractor's delay damages, delay-in-start-up insurance and the sponsor's completion support. It is why they exist.
Breakevens on the same frozen debt: generation can fall 31.8% in every year (beyond one-year P99.9) and the tariff 31.7% (to R0.65/kWh) before cover reaches 1.00x.
The whole pack, 12 cases and 2 breakeven solves: 1.0 seconds.
Next: checking the model.
Riverbend Solar is an illustrative case study: a fictional deal with no real counterparty.