Checking your access…
CiO LabProjection Workbench

Basket cost index over the programme

Monthly spend: at award prices vs projected

UGX million

Where the escalation comes from

contribution by material, median scenario

Material detail

forecast from the last published index month

Programme against the projected price path

bars are phases; colour is the phase's own basket escalation at its mid-point

Phases

edit share, start and duration (percent of programme); baskets follow the template

Clause parameters

%
%
months

The Weighted Macro-Volatility Index is the basket's rolling six-month volatility plus exchange-rate volatility scaled by the basket's pass-through, standardised against its own history. When it breaches the trigger the contract sum is revised towards realised cost, capped per revision, and the revision persists.

Contract sum vs realised cost

Contractor margin through the programme

fixed price vs DAM

Final margin across simulations

distribution of the end-of-contract margin, fixed price vs DAM

Constraints

months early
% / month

Need month for each material is the value-weighted centre of the phases that use it. The finder compares buying at need against buying earlier, net of holding cost, along the chosen forecast path.

Recommended purchase windows

index at need vs best window; savings on the material's share of the contract

How the projections are made

transparent by design; every parameter is in the page
Data. UBOS Construction Input Price Index material series, spliced across rebasings (panel v1.0), monthly to . Months between then and the project start are themselves forecast.
Per-material model. Monthly log-return = drift + calendar-month seasonal + pass-through × shilling return + shock, where the shock has lag-1 persistence φ and volatility σ. Drift and σ are exponentially weighted (24-month half-life) so recent behaviour counts more; the full-sample alternative is selectable.
Joint simulation. Materials share a common market factor (loading from the average pairwise return correlation, ), so basket risk is not understated by treating materials as independent. Percentiles are taken across simulated paths of the basket, not per material.
Scenarios. Base: drift and volatility only. FX shock: an additional depreciation of the shilling spread over the next twelve months, passed through with each material's estimated elasticity. 2022 replay: each material's actual monthly returns from March 2022 to March 2023 are replayed from the project's first month, with noise on top.
Escalation. Contract value is spread over the programme by the spend profile; each month's spend is scaled by the basket index relative to the award month. Escalation is the sum of those differences. Contingency is the escalation at the upper band.
Limitations. Bitumen and fuel are not yet in the panel (diesel stands in for both); labour is UBOS's published series. The model is statistical, not structural — it does not know about policy or supply events that have not appeared in the index yet. Charles's GRU/LSTM forecaster (paper P9) can replace the per-material model by writing the same parameter file.