Separate gauges, not one flow
USD · PER-GAUGE BASISPace of spending
PLAN-TO-PLAN · USDPace of awarding
BY QUARTER · USDAgency velocity
TOP 10 BY PIPELINE| Agency | Open pipeline | Spent (TTM) | Awarded (TTM) | Backlog, years |
|---|
Funnel · Methodology & definitions
SERIES NOTESThe measures
Committed, unspent is the forward pipeline from the “Pipeline” page: total project budgets minus spending to date across active capital projects in the City's most recent capital plan snapshot (NYC OMB, dataset fb86-vt7u). Awarded is trailing-twelve-month construction award value from the City Record (dataset qyyg-4tf5), as on the “Awarded” page. This figure is gross: deduplicated to one record per procurement but still inclusive of renewals, extensions, and awards of $500M and above. It sits on the same basis as the headline on the “Awarded” page and the “Awarded, 12 mo” tile on the home page, which is the site's canonical gross reading. Every new-work reading runs below it: the competitive-set total on the “Contractors” page, the index on the “Analysis” page, and the trailing-year figure in the “Briefing”, each of which strips renewals and megaprojects to isolate contested new demand. Spent is trailing-year capital outlays, measured as the growth in each project's cumulative spend-to-date between capital plan snapshots, across the projects observed in both. The underlying ledger is the City's FMS, the same accounting that feeds Checkbook NYC.
Paid (Checkbook) is a separate reading drawn from the Comptroller's Checkbook NYC: total disbursements on registered construction contracts to named vendors, City scope, fiscal years 2021 to 2025. It is the demand-to-cash payoff at the vendor level, the same figures shown on the “Contractors” page, summed across all construction vendor groups. It is not the same number as Spent. Spent is capital-plan outlays across all construction measured between plan snapshots; Paid is disbursements on the subset of work that has been registered as a contract to a named vendor. The two sit on different ledgers and different scopes and are never added together.
How spending pace is computed
OMB publishes a plan snapshot roughly every four months. For each project, spending in a period is its cumulative spend-to-date in one snapshot minus the prior snapshot, floored at zero. Projects are matched across snapshots by agency and FMS ID, with sub-project rows collapsed first so nothing is double-counted.
Spending is measured only across projects observed in both the prior and the current snapshot, a matched panel. The reason is that spend_to_date is a lifetime cumulative figure, not a period figure. A project appearing in the plan for the first time arrives carrying whatever it has already spent, sometimes years of it; counting that from zero would book its entire back-history as spending in the one period it happened to show up. A new entrant therefore contributes nothing until it has been observed twice, and its spending before it entered the plan is not counted. Projects that leave the dataset stop contributing at the same rule. Both edges hold the series to growth that two readings actually support.
Backlog coverage
Unspent committed dollars divided by trailing-year spending: how many years the current funded pipeline would take to deliver at today's pace. It's the demand-side counterpart to a contractor's backlog ratio. A higher number means more future work stacked behind the same delivery capacity.
Scope
Committed and Spent cover the City's full capital plan (buildings, transportation, water, parks) because capital budgets aren't divided by procurement category. Awarded covers the City Record's two construction procurement categories. Every gauge answers one question from its own ledger: how demand is moving. They are not a single cohort traced end-to-end, and the methodology keeps each measure on its own source's terms.
Agency velocity
Agencies are matched across the two sources (capital plan codes to City Record names) for the agencies that appear in both. Backlog years is each agency's open pipeline divided by its own trailing-year spending, shown only where that spending clears $10M. Below $1M of trailing outlay the spend figure itself is suppressed to an en dash: an agency whose whole year of self-delivered spending lands in the thousands is not spending slowly, it is not the builder, and its capital work is delivered under another managing agency, chiefly the Department of Design and Construction. Printing the residue would read as a figure rather than as the absence of one.