Stress testing

Put a 1-in-100 flood and a delayed transition through your book.

Combine physical shocks with NGFS transition scenarios and see what they do to probability of default, expected credit loss and CET1 capital, with every number traceable to the hazard, the scenario and the data delivery behind it.

Physical shocks
7 types
Transition scenarios
4 NGFS
Outputs
PD · ECL · CET1
Horizons
today · 2030 · 2050 · 2090
ST-0147 · Extreme rain (1-in-100 flood) + Delayed Transition
Horizon 2030 · whole book · reporting currency ZAR
OfficialReviewed
Portfolio PD shift +41 bps P10–P90 +29 to +55
Expected credit loss uplift +R 1.8 bn P10–P90 1.3 to 2.4
CET1 ratio impact −62 bps P10–P90 −44 to −83
Pinned deliveries this run read
  • Hazard indicators · ERA5 1991–2025v12
  • 2D flood depth on terrainv4
  • NGFS transition pathwaysv8
  • Counterparty book · Q3v3
runST-0147 · officialZARIllustrative
Seven physical shocks

Shocks anchored in what has actually happened.

Each physical shock is defined against observed history or a named return period, so “severe” means something you can point to: the hottest year in a place’s record, a drought like a year your credit committee remembers, the flood depth expected once in a hundred years.

01 · Extreme temperature

Heatwave like a record year

Heat at the intensity of the worst year in each place’s observed record, applied to every site in the book at once.

ERA5 1991–2025
02 · Water

Drought like a chosen year

Pick a drought year from the record and replay it across the book, measured with the Standardised Precipitation Index.

SPIobserved year
03 · Water

Extreme rain with a 1-in-100 flood

Rainfall at a chosen return period and the flood depth it produces at each site, from 2D modelling on bare-earth terrain.

1-in-1002D flood
04 · Fire

Fire season

A severe fire season expressed through the fire risk score at each site, for agricultural, forestry and peri-urban assets.

fire risk score
05 · Geotechnical

Seismic event, 1-in-475

Peak ground acceleration at the 1-in-475 year level, the reference return period used in seismic design.

PGA 1:475
06 · Geotechnical

Landslide in a wet season

Landslide susceptibility under a wet season, where slope, soil and rainfall meet on the terrain model.

susceptibilityDTM
07 · Extreme temperature

Heat discomfort

Sustained heat and humidity that slow outdoor work and raise cooling loads, measured with comfort and humidex indicators and WBGT.

humidexWBGT
+ Transition

Any of four NGFS scenarios

Net Zero 2050, Delayed Transition, Current Policies or NDCs, turned into carbon cost, revenue and credit impact for each counterparty.

4 NGFScarbon cost
Combine and calibrate

One physical shock, one transition path, one horizon.

A scenario is a physical shock at a chosen severity, an NGFS transition pathway and a time horizon. Physical and transition risk are kept as two dimensions all the way through, so the result shows which one is doing the damage.

  • Severity is set against observed history or a return period, not an arbitrary multiplier.
  • Horizons at today, 2030, 2050 and 2090, with CMIP6 SSP projections behind the physical hazard.
  • Run the same shock under several transition pathways and compare them side by side.
  • Results in your organisation’s reporting currency, such as ZAR, KES or NGN.
New stress test: scenario definition
Exploratory draft
Physical shock HeatwaveDroughtExtreme rainFire seasonSeismicLandslideHeat discomfort
Severity Like 1992Like 2015Record year per place
combined with
NGFS pathway Net Zero 2050Delayed TransitionCurrent PoliciesNDCs
Horizon Today203020502090
Book Agricultural lending · 1,284 counterparties
shockSPI drought · ERA5SSPCMIP6Illustrative
The results chain

From the water at the gate to the line in the capital plan.

Every stage is stored and inspectable, not just the final number. Here is one counterparty, CP-0417, under ST-0147: open any link in the console to see the inputs, the range and the version it came from.

  1. 01 · Hazard

    Flood depth at the site

    The shock is resolved at each site: here, depth at the 1-in-100 year level from 2D modelling on terrain.

    1.1 m range 0.8–1.4 m, site S‑2207
  2. 02 · Exposure

    What sits in the water

    Sites are matched to the counterparties and facilities in your book that depend on them.

    R 240 m drawn exposure · 2 facilities
  3. 03 · Counterparty

    Revenue and carbon cost

    Physical and transition effects on the counterparty, kept apart so each can be read on its own.

    −7.5% revenue + R 14 m carbon cost a year
  4. 04 · Default

    Probability of default

    The weaker financial position moves the counterparty’s PD and, where it crosses a band, its severity.

    1.8% → 3.1% At risk → Stressed
  5. 05 · Loss

    Expected credit loss

    PD meets exposure and loss given default to give the uplift in expected credit loss.

    +R 6.2 m range 4.4–8.3 m
  6. 06 · Capital

    CET1 contribution

    Loss flows through to the CET1 ratio, summed across the book with every counterparty’s share visible.

    −0.4 bps of −62 bps for the book

All values for CP-0417 and site S-2207 are synthetic Illustrative

PD, ECL and CET1

See how much of the capital hit is physical, and how much is transition.

Expected credit loss and CET1 impact are reported with a range, not a point, and split by source. A board can see that a scenario costs 62 basis points of CET1 and that 38 of them come from the flood.

  • Probability of default before and after the shock, for each counterparty and in aggregate.
  • Expected credit loss uplift by sector, region and facility.
  • CET1 ratio impact as a waterfall from baseline to stressed.
  • Severity bands that always carry a word, never colour alone.
CET1 ratio, baseline to stressed, ST-0147, 2030
13.58 % stressed
from 14.20%
−62 bps (−44 to −83)
14.4%14.0%13.6%13.2% 14.20 −38 −24 13.58 BaselinePhysicalTransitionStressed
runST-0147 · officialbpsCET1 ratioIllustrative
Runs you can stand behind

Explore freely. Publish deliberately.

Analysts need room to try things; supervisors need to know which number is the real one. ecosphere keeps the two apart, and pins every run to the exact data it read so the result can be reproduced next quarter.

Exploratory

For working it out

Try a different severity, swap the transition pathway, narrow the book to one sector. Exploratory runs are clearly labelled wherever they appear and never reach a report by accident.

  • Any analyst can create one
  • Labelled “exploratory” on every figure and export
  • Comparable side by side with official runs
Official

For the committee and the supervisor

An official run is the one your institution stands behind. It reads pinned deliveries, is reviewed by a second person, and is the only kind that exports as a scenario report of record.

  • Pinned to specific delivery versions
  • Reviewed by someone other than its author
  • Exports as a scenario report

Pin a delivery

A run records the version of every delivery it read: hazard indicators, flood depths, NGFS pathways and your own counterparty book. When a newer delivery is published, the pinned run does not move until you choose to rerun it.

Trace every result

Any figure, from a single PD to the CET1 total, links back through the results chain to the deliveries it was computed from, with their source, version and reviewer.

Export the scenario report

The scenario report sets out the scenario definition, the deliveries read, results by portfolio and counterparty, and the ranges around them, ready for risk committee and supervisory review.

Questions it answers

The questions risk committees actually ask.

  1. If the 1992 drought happened again, how much would our expected credit loss rise?Replay an observed drought year across the agricultural book and read ECL uplift by region and sector.
  2. How much CET1 would a delayed transition cost us by 2030?Run Delayed Transition alone, then with a physical shock, and compare the capital impact side by side.
  3. Which counterparties move into Stressed, and is it physical or transition risk?Both dimensions are kept separate, so every band change shows its cause.
  4. What does a 1-in-100 flood do to the facilities secured on riverside sites?Flood depth at each site carries through exposure and PD to loss, per facility.
  5. Is this the same number we reported last quarter?Official runs are pinned to delivery versions, so you can reproduce last quarter’s result exactly and see what changed since.
  6. Can we show the supervisor where each figure came from?The scenario report and the results chain trace every number back to its source, version and reviewer.
See it on your own book

Bring a sample portfolio. Leave with a stress result.

In a working session we run a sample of your counterparties through a physical shock and a transition scenario, and walk you through every step of the chain from hazard to capital.