U.S. federally funded hazard mitigation construction, 1999-2024

These projects rarely blow their budgets. They quietly deliver less, and late.

An interactive companion to a portfolio analysis of 14,906 construction projects funded through FEMA Hazard Mitigation Assistance grants: buyouts of flood-prone homes, elevations, safe rooms, drainage and more. Built entirely from public OpenFEMA transaction records and the NOAA storm archive.

The funding outcome

Under-delivery, not cost overrun

Grant-funded construction cannot overrun the way contract work does: the award is a ceiling. What the obligation ledger shows instead is money handed back. More than twice as many projects finish materially below their funded amount as above it.

Where projects end up against their award

Final obligation ÷ initial award, share of projects
Descoped >5% · 43.2%Held ±5% · 38.8%Grew >5% · 18.0%

Who gets money taken back

Share of projects with at least one deobligation, by activity type
The activities that depend on homeowners still saying yes years later, buyouts above all, shrink most. Equipment purchases hold their funding.
The schedule outcome

Late against their own promise

Where applicants had to propose a schedule to win a competitive award, 92% blew past it, at a median of 2.59× the proposed duration. The benchmark is not an analyst's baseline. It is the number the applicant wrote to win the money.

Schedule growth by grant program

Realized ÷ proposed duration: median dot, P25–P75 band, P90 tick
Programs whose applicants promise the shortest schedules grow the most.

Median delivery time by approval year

Years from approval to closeout, closed projects, 1999–2021
A slow drift downward of roughly 0.44 years per decade. Recent cohorts exclude still-open projects.
The administrative tier

Where a project lives decides how fast it moves

About 30% of the variance in delivery time sits between administering state offices, not between projects. The same kind of project averages 2.3 years in the fastest states and over 8 in the slowest. Click a state.

Mean years to closeout, by administering state

Empirical-Bayes shrunk estimates; states with ≥25 projects
darker = slower
The implication, as a tool

Schedule reality check

Proposed schedules in competitive applications are commitment devices, not forecasts. This tool applies the observed growth distribution for your activity type to your proposed duration, the reference-class forecast the portfolio itself implies. Planning numbers, not a guarantee.

What history says about a proposal like yours

median realistic duration
middle 50% of comparable projects
Method and data

What this is built on

Sources

OpenFEMA HMA project register56,114 projects
Financial transaction ledger139,378 transactions
Subapplications (proposed schedules)62,791 records
NOAA Storm Events archive1.3M county events
Retrieved July 2026. All statistics on this page are precomputed aggregates of the public files, stored as JSON in the data folder; no individual project data is served.

Reading notes

Delivery time runs from initial federal approval to closeout and is estimated with censoring retained, so still-open projects are counted rather than ignored. Award size is the first award transaction in the reconstructed ledger. Deobligation is a formal federal action removing obligated funds; it is measured here as recorded and cannot by itself separate thrift from abandoned scope. Schedule growth covers the six competitively awarded programs, the only ones that record a proposed period of performance.