Government-lending monitor: FHA, VA and USDA in Ginnie Mae's May 2026 disclosure

Ginnie Mae's single-family disclosure files are the most granular public window onto the government-insured mortgage market — the FHA, VA, USDA and PIH loans that sit outside the conventional GSE channel. This monitor reads the May 2026 disclosure file at the loan level: 154,562 loans, $52.6B in unpaid principal balance (UPB), spread across 190 issuers. Treat it as a cross-section — a monthly snapshot of what the file captured, not a claim about the full outstanding book or a clean measure of new issuance.

The program mix: FHA by count, VA by balance

The government-lending split in this file is the familiar one, and the divergence between count and balance is the point worth holding onto.

• FHA dominates by loan count: 91,266 loans, or 59% of the file, carrying $28.0B of UPB (53% of the total).
• VA is second by count at 59,617 loans (39%) but punches above its weight on balance — $23.8B, or 45% of UPB.
• USDA Rural Development is the small rural slice: 3,525 loans and roughly $0.7B (~2%).
• PIH / Native American (Section 184) is a rounding line here at 154 loans.

The asymmetry is the story. VA supplies 39% of the loans but 45% of the balance, which means the average VA loan in this file is materially larger than the average FHA loan. That tracks with what we know about the two programs: VA's no-down-payment structure and its concentration among borrowers purchasing at or near full entitlement pull average balances up, while FHA's lower loan limits and its role serving lower-balance, higher-LTV first-time buyers pull the other way. For anyone modeling prepayment or convexity across a Ginnie pool, that program-weighted balance gap matters — VA and FHA cohorts do not behave identically, and their relative share moves the aggregate.

Issuer concentration is high

Across the 190 issuers in the file, holdings are far from evenly distributed. The top four issuers hold 41.4% of UPB, and the single largest holds roughly 12.5% — about $6.6B on its own. A concentration curve that steep at the top is consistent with the well-documented structural shift in the Ginnie market over the past decade: the migration of government servicing and issuance away from depositories and toward large nonbank originator-servicers.

That concentration is not a neutral fact. Ginnie issuers carry the operational obligation to advance and pass through principal and interest and to buy out seriously delinquent loans; the guarantor's exposure runs through them. When 41% of a pool sits with four counterparties, servicing-transfer risk, advance-financing liquidity, and issuer solvency stop being idiosyncratic and start being portfolio-level questions — which is precisely why nonbank concentration remains a live thread in the policy conversation around issuer capital, liquidity, and eligibility standards.

One important limit: this file exposes issuers by numeric ID, not by name. We can measure how concentrated the book is, but naming the largest holders requires a separate ID-to-name resolution step against Ginnie's issuer registry, and we are not guessing at identities here.

What the file will not tell you

• One monthly file — a point-in-time cross-section, not a full-year flow or the active universe.
• Underwriting fields suppressed — no LTV, DTI, or MSA on the single-family records, so credit-box and geographic cuts need other sources.
• Issuers by numeric ID — every concentration figure above is name-agnostic until resolved against Ginnie's registry.

Used within those edges, the file does exactly one thing well and does it cleanly: it sizes the government-lending market by program and shows how tightly the book clusters among a small set of issuers — a starting frame for counterparty, servicing, and policy questions, not the last word on any of them.