SA CPI Commodity Pass-Through Model

Column definitions

• Elasticity — the assumed pass-through coefficient (0–1): the share of a commodity's USD/ZAR price move that actually reaches the mapped CPI item's retail price, after retailer margin absorption, regulation, and consumer substitution are netted out. 1.0 = full 1-for-1 pass-through (e.g. regulated fuel pricing); 0.3 = only 30% reaches the till, with the rest absorbed elsewhere in the supply chain (e.g. gold price into jewellery retail prices). This is a judgement input, not an estimated regression coefficient — treat it as a starting point to argue with.
• Indirect uplift (Indirect×) — a manual multiplier scaling the direct effect up to proxy for second-round effects: diesel → fertiliser/animal feed → food; diesel → backup power/generation costs. "Flagged, not modelled" means this number is a guessed scaling factor, not a derived figure — a genuinely modelled version would require tracing each link in the chain separately, each with its own lag and elasticity (e.g. a distinct "diesel → fertiliser → maize" row with a longer lag and lower confidence than the direct diesel → fuel-price link). The multiplier approach keeps the tool simple and honest about its limits, at the cost of not showing where the extra impact is coming from. If a specific pathway (diesel-linked effects are the most material candidate) needs to carry real weight in a client conversation, it's worth breaking out as its own explicit row rather than relying on the hidden multiplier — flag this if useful and it can be added.
• Confidence — reflects reliability of the pass-through assumption itself (data quality, historical consistency), not confidence in the underlying price forecast.
• Lag (mo) — months before the price move is fully reflected in the printed CPI index. Shifts timing of impact along the chart's monthly path; does not change magnitude.

How the CPI bps figures are calculated

• For each commodity: ZAR price change = (futures-implied or what-if USD % change) + (FX forward % change for that tenor).
• Direct CPI impact (bps) = CPI sub-weight% × elasticity × ZAR price change% × indirect uplift multiplier × 100.
• Lag (months) shifts when the impact lands in the printed index — it does not change the magnitude. The chart applies lags when spreading impacts across the monthly path.
• Fuel benefits from a known, short, mechanical lag (SA's monthly retail price-averaging + one-month publication lag). Food/agricultural pass-through is longer and considerably noisier — local harvest cycles, import parity, and retailer margin absorption all interfere — hence lower confidence and wider assumed lags.
• SA Maize, SA Soybeans and SA Wheat are priced off SAFEX (JSE Commodity Derivatives) rather than CBOT — these contracts settle in ZAR, so the model does not add the FX overlay on top of their spot/futures % change (it's already embedded in the local price via import/export parity arbitrage). Their assumed lag is shorter than the original CBOT proxies, reflecting the absence of a shipping/FX-conversion lag on top of the local harvest cycle.

What is explicitly not modelled (guesstimates only)

• Second-round effects (diesel → fertiliser/agri input costs → food; diesel → backup power/electricity generation costs) are captured only via the manually-set Indirect Uplift multiplier — this is a judgement input, not a derived figure.
• Retailer/producer margin absorption — in practice, price moves are frequently partially absorbed rather than fully passed through, especially on the way down.
• SARB reaction function / monetary policy feedback loops.
• Cross-commodity substitution effects (e.g. consumers shifting between beef and poultry) and demand elasticity.

Reference weights used as starting points

• BCOM constituent weights: Bloomberg Commodity Index factsheet, 30 June 2026.
• SA CPI main-category weights (2023 basket, current at time of writing): Housing & Utilities 24.10%, Food & NAB 18.23%, Transport 13.89%, Insurance & Financial Services 10.41% — Stats SA basket/weights update, January 2025.
• Fuel, cereal, meat and oils/fats sub-weights are Stats SA-sourced where available; where not, they are flagged "Est." and should be checked against the latest COICOP 8-digit weights file before client use.