How Often a Declared County Is Actually Covered for Household Assistance

Executive Summary

A county appearing in a federal disaster declaration is not the same thing as a county whose households can apply for federal help. FEMA's own explainer keeps the household-facing and government-facing programs separate, and its designated-area listings report them one declaration at a time.

What this paper measures, and what it does not. It measures which designation-level program flag FEMA recorded for a county in a declaration. It does not measure applicant approval outcomes, and it never inspects an applicant record. That distinction matters because CountyGuard already publishes research on FEMA approval rates among applicants; this paper sits upstream of that question, at the point where a county is or is not designated for a household program at all.

This paper measures how often the two coincide, using the federal declaration record itself. Across 8,442 county designations in 605 declarations covering Colorado, Oklahoma and Texas, 30.5% carry one of FEMA's two household-assistance flags. The number on its own is not the finding. The finding is compositional: the household-flag share differs sharply by declaration type. A county named in a major disaster declaration carries a household-assistance flag 50.2% of the time, a county named in an emergency declaration 0.07% of the time, and a county named in a fire management declaration never.

That composition is the reference contribution. A reader who knows the declaration type already knows most of what the flag will say, and a reader who knows only that "my county was declared" knows very little.

The most important limitation is what the record does not contain. The governed table carries FEMA's iaProgramDeclared and ihProgramDeclared fields and no public-assistance field at all. A county without a household flag therefore cannot be described as covered for government-facing programs only; the paper can say what is present and must stay silent on what is absent.

The practical implication is bounded and specific: check the program flags for your own county and disaster rather than inferring eligibility from a declaration headline. This is a reference benchmark. It names no action for any agency and no entitlement for any household.

Business Question

Question. Among counties named in federal disaster declarations in Colorado, Oklahoma and Texas, how often does the designation actually carry a household-assistance program flag, and how much does that share vary by declaration type and incident type?

Audience. Residents, local journalists, civic organisations and county staff who read public-record disaster declarations — the local public reader.

Publication-value mode. Reference benchmark. The evidence creates a durable, reusable comparison that a simple authoritative-table lookup does not supply: FEMA publishes designated-area listings one declaration at a time, and its interactive cross-declaration visualisation was withdrawn in favour of a raw data interface.

Specific reader utility. A reader can establish, for the represented period, how often each declaration type carried a household-assistance flag, and can use that observed composition when reading an individual announcement instead of inferring eligibility from the headline.

Positive takeaway. If household coverage is uneven, a resident has a concrete reason to check the program flags for their own county rather than infer eligibility from a declaration headline.

Credible null takeaway. If household coverage were near-uniform inside comparable declaration classes, the recurring public confusion would be a communication failure rather than a coverage pattern, and effort would belong with how designations are announced.

Why it matters now. In May 2026 a federal review council recommended raising disaster declaration thresholds through updated per-capita indicators and new minimum state expenditure requirements, with estimates of roughly sixteen fewer major disaster declarations a year, and recommended replacing the current Individual Assistance structure with a single payment. The observed share of designated counties carrying a household program is the quantity that debate rests on, and it has not been published at this grain.

Data & Scope

Source. FEMA Disaster Declarations Summaries, as represented in the governed Eldhurst research data platform. FEMA's population noun is the designated area; this paper's unit is one designated county within one declaration.

Universe. The governed declaration table covers Colorado, Oklahoma and Texas. All results are bounded to those three states and are not national.

Grain. One unique pair of disaster number and county FIPS code. The statewide designation table is a separate governed table and is excluded entirely.

Period. All declaration incident dates represented in the governed universe, with incident begin dates spanning 1965 to 2026.

Inclusion rule. A row is included when county_fips is not null, which is the governed measure's mandatory filter.

Exclusions and their counts. Zero rows were dropped for a null county code. Zero duplicate (disaster_number, county_fips) pairs were collapsed. Zero pairs carried inconsistent declaration or incident type. Zero rows carried a null value in either program flag. Nothing was silently dropped.

Fields. ih_program and ia_program are FEMA's ihProgramDeclared and iaProgramDeclared. They are separate fields and this paper never treats them as interchangeable. Every share is reported for each field individually and, where a union is quoted, the union is named explicitly as "either flag set".

Coverage gap that shapes every claim. The governed table carries no public-assistance flag and no hazard-mitigation flag. The absence of a household flag on a designation is therefore not evidence that the county received a different program. It is the absence of a record this dataset does not keep.

Source limitations disclosed by the publisher. FEMA describes older declaration records as manually entered, and fire-management records as historically partial. The era series in Section 5 must be read with both in mind.

Method

The calculation is a share. For any group of county designations, the household-assistance share is the number of designations with ih_program set, or with ia_program set, or with either set, divided by the number of designations in that group. Each of the three is reported separately; none is derived from another.

Designations are deduplicated to one row per disaster number and county code before any share is computed, taking the maximum of each flag so that a county recorded more than once for one disaster cannot be counted twice or lose a flag.

Groups are formed by declaration type, then by declaration type crossed with incident type. A crossed cell is reported as meeting the minimums when it holds at least 5 distinct disasters and at least 50 county designations; ten cells qualify. Cells below the floor are published in full but are marked as not meeting the minimums rather than being hidden or merged.

The two locked sensitivities restrict incident begin dates to 2000 onward and to 2015 onward, both fixed before any outcome was seen. Against the full-period household-flag share of 30.5023%:

Every share and shift is computed from integer counts and rounded once, at serialization. The 2015-onward shift is abs(903/2692 - 2575/8442).

A separate six-band view by era of incident begin date is reported in Section 5; it is a descriptive series that exposes the field transition, not one of the two locked sensitivities.

Composition decomposition. The locked failure clause asks whether the household share moves by more than 0.25 between the full period and the 2015-onward restriction without that movement being explainable by declaration-type composition. The first operand is evaluated first: the observed shift is 0.030416, far below 0.25, so the clause cannot fire and the explanation operand is never reached.

The change is nonetheless decomposed, as a descriptive result, into a composition component and a within-declaration-type rate component using a Kitagawa symmetric additive decomposition.

The components sum exactly to the observed change, and the identity is asserted at run time. Both one-sided counterfactuals are also emitted. These figures describe how the observed change splits between a changing mix of declaration types and changing rates within them. They are descriptive only. No rule converts them into a judgement that composition does or does not explain the movement, because no such rule is authorized by the locked specification, and one may not be introduced after the outcome is known.

The robustness condition is therefore false on its first operand, and composition_explanation_evaluation is recorded as NOT_REACHED.

There is no ranking rule and no decision rule. The paper produces no ordered list of counties and names no county.

Full implementation detail, field definitions and the reproducibility command are in Section 9.

Findings

The locked evidence artifact: composition by declaration type

Figure 1 renders the "either flag" column.

Across all 8,442 designations, 284 (3.4%) carry both flags. The three declaration types behave almost as three different records: about half of major-disaster county designations carry a household flag, 2 of 2,894 emergency designations do, and none of the 426 fire-management designations does.

Composition holds when incident type is added

Ten of the crossed cells meet the minimums.

Inside major disasters the share still moves — from 25.0% for fire to 68.5% for tornado — but the gap between declaration types is larger than the gap between incident types inside any one of them.

Within major disasters, by state

This is observed composition across three states with very different declaration histories. It is not a comparison of how comparable counties were treated, and it supports no ranking.

The two flags are used in different eras

Figure 2 renders this table.

The last incident year carrying an iaProgramDeclared flag is 2008; the first carrying ihProgramDeclared is 2002; both appear in 2002, 2003 and 2005 through 2008. Restricting to incident begin dates from 2015 onward gives 2,692 designations with an ihProgramDeclared share of 33.5% and an iaProgramDeclared share of exactly zero.

This is a change in which field FEMA populates, and nothing else. A reader who took the modern zero in iaProgramDeclared as evidence that Individual Assistance stopped being declared would be reading a schema change as a policy change. The two fields must be read as one household-facing signal across the whole period and as two separate fields within any single era.

Interpretation, kept separate from the above

The direct findings are the shares in the tables. The interpretation is this: the declaration record answers a narrower question than the public announcement implies. "Your county is in the declaration" and "your household can apply" are different statements. Across the represented period they coincide for about half of major-disaster county designations, for two of 2,894 emergency designations, and for none of the fire-management designations.

Association Analysis

No inferential association test is run, and none is justified by the locked method.

The locked method for this paper specifies descriptive shares at a fixed grain, a declaration-type decomposition and an era sensitivity check. It authorises no significance test, correlation or model, and this section does not add one.

The reason is not caution about the result; it is that the universe is not a sample. These are all represented county designations in the governed universe, not a draw from a larger population. A p-value or confidence interval computed here would describe sampling variability that does not exist and would invite a reader to generalise beyond Colorado, Oklahoma and Texas, which Section 8 forbids.

The prespecified checks that were authorised are reported in Sections 4 and 5: the decomposition by declaration type and incident type, the two locked sensitivities at 2000 onward and 2015 onward, and the descriptive Kitagawa decomposition of the 2015-onward change. The declaration-type result is descriptive: household-flag shares differ across declaration types, and most designations carrying a flag are major-disaster designations. That is a statement about the composition of the record, not an association between independent variables and not an explanation of it, and it is stated as such.

Association is not causation, and in this paper no association is claimed at all.

Business Implications

The declared value mode is a reference benchmark, so the permitted implication is what a reader can now understand, compare and check — not an action anyone must take.

A resident or local reporter can:

• read a declaration announcement and know how often, across the represented period, each declaration type carried a household-assistance flag: about half of major-disaster county designations, 2 of 2,894 emergency designations, and none of the fire-management designations;
• check the program flags for their own county and disaster rather than infer eligibility from a headline;
• treat 50.2% as the observed share for major-disaster county designations in these three states across 1965-2026, not as a forecast for a new declaration. The field-usage shift described in Section 5 means the pooled historical share is not a governed estimate of current practice, and this paper does not supply one.

Anyone using the federal record analytically can:

• treat ihProgramDeclared and iaProgramDeclared as one household-facing signal across the full period and as two distinct fields within an era;
• avoid building a time series from either field alone, which would show a spurious collapse to zero in 2008 or a spurious appearance from nothing in 2002.

What must not be inferred or automated. Nothing here supports an eligibility determination for any household, any statement about physical damage or unmet need, any claim about requests or denials, any assertion that a county deserved or did not deserve a designation, or any inference about counties outside Colorado, Oklahoma and Texas. No county is named or ranked. Because the governed record carries no public-assistance field, no designation may be described as covered for government-facing programs only.

Limitations

What the data exclude. The governed table carries no public-assistance flag, no hazard-mitigation flag, no request that was denied, no physical damage measure and no measure of unmet need. It is an administrative record of approved declarations and designations.

What the method cannot determine. Whether any household received assistance, whether a county should have been designated, whether comparable counties were treated differently, and whether the same composition holds outside the three represented states. The paper measures which flag a designation carries and nothing about the process that produced it.

Coverage and sensitivity weaknesses. FEMA describes older declaration records as manually entered and fire-management records as historically partial. The 1965–1979 and 1980–1989 bands rest on 471 and 296 designations respectively, the two smallest in the series, and the very high household shares there should be read as a property of a small, old, manually entered record rather than as evidence that coverage was once near-universal. Nine crossed cells fall below the minimums and are published but marked.

Alternative explanations considered. The strongest is that the flag pattern follows from declaration type, because the household programs are associated with major-disaster declarations rather than with emergency or fire-management ones. The observed composition is consistent with that explanation, and this paper reports the composition rather than testing the mechanism: the governed record carries the flags and the declaration type, and nothing about the statutory process that set them. The consequence is that the result must be read as the composition of the declaration record, never as variation in treatment.

The unanswered question that should guide follow-up. This record shows which program flag a designation carries. It cannot show how many counties requested a designation and did not receive one, because requests and denials are outside the governed universe. That is the question a reader will ask next, and it needs a different source.

References / Appendix

Authoritative sources

• Federal Emergency Management Agency, OpenFEMA Dataset: Disaster Declarations Summaries – v2 — defines iaProgramDeclared and ihProgramDeclared as separate fields, the three declaration types, designated areas, and the historical-data limitations.
• Federal Emergency Management Agency, Individual Assistance vs. Public Assistance — distinguishes household-facing assistance from the government-facing program.
• Federal Emergency Management Agency, Designated Areas — the per-disaster listing a reader is routed to today, which answers one declaration at a time.
• National Association of Counties, FEMA review council releases final report recommending sweeping changes, 11 May 2026 — the declaration-threshold and Individual Assistance restructuring proposals that make the observed share timely.

All external sources were verified on 12 August 2026.

Field definitions

Permitted and prohibited claims

This article states: the shares in Sections 5 and 7, stated as composition of the represented declaration record for Colorado, Oklahoma and Texas.

Outside what this article claims: describing the two FEMA fields as interchangeable; inferring any program absent from the record, including any public-assistance designation; implying variation in how comparable counties were treated; inferring physical damage, unmet need, a request, a denial or entitlement; treating the post-2008 zero iaProgramDeclared value as evidence that household assistance disappeared; and generalising beyond the three represented states.

Figures

Scroll chart horizontally to read all labels.

Full chart description

Horizontal bar chart titled "Household-assistance coverage by declaration type". It shows the share of designated county records in Colorado, Oklahoma and Texas that carry FEMA's iaProgramDeclared or ihProgramDeclared flag, for each of three declaration types. - Major disaster (DR): 50.2% of 5,122 county designations. - Emergency (EM): 0.1% of 2,894 county designations. - Fire management (FM): 0.0% of 426 county designations. The chart shows composition of the declaration record. It does not show physical damage, need, requests, denials, entitlement, or differences in how comparable counties were treated.

Scroll chart horizontally to read all labels.

Full chart description

Grouped horizontal bar chart titled "Which household-assistance field is populated, by era". For each era it shows two bars: the share of designated county records with ihProgramDeclared set and the share with iaProgramDeclared set, in Colorado, Oklahoma and Texas. - 1965-1979: ihProgramDeclared 0.0%, iaProgramDeclared 86.0%, across 471 county designations. - 1980-1989: ihProgramDeclared 0.0%, iaProgramDeclared 77.0%, across 296 county designations. - 1990-1999: ihProgramDeclared 0.0%, iaProgramDeclared 35.5%, across 1,179 county designations. - 2000-2009: ihProgramDeclared 9.6%, iaProgramDeclared 17.2%, across 3,081 county designations. - 2010-2019: ihProgramDeclared 22.5%, iaProgramDeclared 0.0%, across 1,417 county designations. - 2020-2026: ihProgramDeclared 33.1%, iaProgramDeclared 0.0%, across 1,998 county designations. The two fields are separate FEMA flags. A field falling to zero shows which field FEMA populated in that era; it is not evidence that household assistance stopped.