What the series counts
Every tender published on Etimad is grouped by the month it was published,
not the month it was awarded. The value shown is the winning bid of those
that have since been awarded. Grouping by publication is what makes this a
read on commissioning tempo rather than on paperwork throughput.
Why recent months are incomplete
A tender is awarded months after it is published, so the newest months are
always understated and revise upward. About 62% of published tenders are
eventually awarded, and the large ones settle last — a month takes far longer
to reach its final value than the three to six months the sector assumes. That
is why the headline above is an estimate rather than a count.
How much later depends on how the work was bought. Two months after
publication, direct purchases have about 41% of their eventual value confirmed
and open competitions have 2%. Reading a recent month by what has landed in it
therefore describes the order the paperwork arrives in, not the market.
How the estimate is built
Each month is anchored on the tenders it published — a count that is known the
moment the month ends and never revises. The estimate is what has been confirmed
so far, plus what those published tenders are still expected to be worth. It is
the Bornhuetter-Ferguson method from insurance reserving: it leans on how big
the month was while the month is young, and converges on the observed figure as
awards land. It never divides by the completion share, which is what stops a
single early award in a fresh month implying a record-breaking total.
A published tender is not priced at the market average. It is priced by the
agency that published it and the route it was bought through, from what that
agency's tenders have historically been worth once settled. Most agencies
publish too little for their own history to be trusted, so each one is weighted
against the market average in proportion to how much evidence it has — the
credibility weighting used in insurance pricing. Each route is then developed on
its own curve, and the month is the sum of its parts.
Why the split is estimated too
Selecting a column decomposes it by route and by sector. Those shares are
estimated on the same basis as the bar, not counted from what has settled — a
counted split would report every recent month as mostly direct purchase, which
is true of the tenders and false of the money. Direct purchase is roughly 65% of
tenders published and 1–2% of value awarded.
Sectors come from Etimad's own activity classification, one per tender. The six
largest by estimated value are named and the remainder is pooled, so the parts
always sum to the month above them.
What the two comparisons rest on
The record begins in July 2025, so only July has a counterpart twelve months
earlier — that single month is the year-over-year reading beside the latest
month, and it sets an estimate against a settled figure.
The year-to-date figure is the sum of the estimates for the months of the
current year so far. It has no like-for-like counterpart yet: 2025 is on record
only from July, so setting seven months of 2026 against six of 2025 would report
a difference in window length as a fall in spending. Until the series covers the
same months in both years, the comparison divides each side by its own length
and compares monthly pace — which is why it is labelled "a month". Procurement
is seasonal, so that is a weaker reading than a like-for-like one, and it
upgrades on its own once the history reaches back far enough.