Macro scenarios — forward-looking overlay

IFRS 9 demands that ECL reflect 'reasonable and supportable' forward-looking information. The macro scenarios page is where that forward-looking lens is built — usually three scenarios (base, upside, downside), each with macro inputs and a probability weight, blended into a single ECL number.

Why macro scenarios exist

Historical default rates tell you what borrowers did in the conditions of the past. IFRS 9 wants ECL to reflect what borrowers will do in the conditions of the future. The bridge between the two is a macro model: a set of macroeconomic factors that explain default behaviour, projected forward under multiple scenarios, with each scenario probability-weighted.

We use three scenarios — a base case (most likely), an upside (better than expected), and a downside (worse than expected). Each scenario adjusts the through-the-cycle PD by a scalar derived from its macro factor projections. The final ECL is the probability-weighted ECL across the three scenarios.

The macro factors we use

Five Kenyan macroeconomic factors drive the PD overlay. They were selected from a longer list after back-testing against the internal default series. Each factor is sourced from a public release and refreshed before each quarter-end.

How factors become a PD overlay

Each segment has a fitted linear overlay equation that maps macro factor deviations from baseline to a PD multiplier. The equations were fitted on 8 years of macro × default data using a least-squares regression.

The floor and cap prevent the overlay from producing unreasonable numbers when a factor moves outside the calibration range. They are documented in the model documentation and the cap has been hit twice in the last three years (Covid-19 H1 2020 and the 2022 FX shock).

# Example — Business SME segment overlay
# Calibrated coefficients (stored on ifrs9.segment.overlay)
#   beta_gdp = -0.45     (1pp lower GDP → +45% PD)
#   beta_cpi = +0.12     (1pp higher CPI → +12% PD)
#   beta_fx  = +0.08     (KES/USD +1% → +8% PD)
#   beta_cbr = +0.18     (CBR +1pp → +18% PD)

def pd_multiplier(segment, scenario):
    s = scenario.macro_factors  # dict of projected values
    delta_gdp = s['gdp'] - segment.baseline_gdp
    delta_cpi = s['cpi'] - segment.baseline_cpi
    delta_fx  = (s['fx']  / segment.baseline_fx)  - 1.0
    delta_cbr = s['cbr'] - segment.baseline_cbr
    raw = 1.0 \
        + segment.beta_gdp * delta_gdp \
        + segment.beta_cpi * delta_cpi \
        + segment.beta_fx  * delta_fx \
        + segment.beta_cbr * delta_cbr
    return max(0.4, min(raw, 2.5))  # floor 0.4×, cap 2.5×

Scenario weights

After each scenario produces its own ECL, the three are blended:

Default weights are 50% base / 25% upside / 25% downside. The Risk Committee can shift weights when forward-looking conditions justify it — most commonly raising the downside weight in periods of macroeconomic stress.

ecl_weighted = w_base * ecl_base + w_upside * ecl_upside + w_downside * ecl_downside
# w_base + w_upside + w_downside == 1.0

Projection horizon

Macro factors are projected over a horizon that covers the longest expected lifetime in the loan book — typically 5 years. Years 1-2 are the 'reasonable and supportable forecast' period (from named institutional forecasts). Years 3-5 revert linearly to the long-term mean.

What you see on the screen

IFRS 9 → Configuration → Scenarios lists the current scenario set. The screenshot shows base / upside / downside with their respective weights, macro factor inputs, and a 'last approved' stamp. Editing requires Risk Manager privileges; posting requires CFO sign-off.

Worked scenarios

Reference

ifrs9.scenario fields

Default shipped scenario set

Troubleshooting

See also

• Ifrs9
• Pd Lgd Ead
• Quarterly Run
• Audit Pack
• Treasury