Guide

Agency bonds explained

Harbor Capital's fixed-income policy said “government quality only” — so the sleeve held T-notes and a sliver of TIPS. Callable agency debentures from Fannie Mae and Freddie Mac traded 35–50 basis points above comparable Treasuries with similar duration, but the investment committee treated them as “quasi-corporate” and excluded them. When spreads widened in early 2025 after a regional-bank scare, IG corporates sold off harder than agencies while Treasuries rallied — the portfolio captured flight-to-quality in bills and notes but missed the middle: agency debt that behaved more like government paper than BBB corporates. Income trailed the liability-matching benchmark by 42 bps for the year.

Agency bonds are debt securities issued by U.S. government-sponsored enterprises (GSEs) and related agencies. They sit between Treasuries and investment-grade corporates on the risk spectrum: not full-faith-and-credit U.S. obligations (except Ginnie Mae), but historically supported by an implicit (“too big to fail”) federal backstop that markets price as very low credit risk. This guide covers issuer and guarantee taxonomy, debentures vs mortgage pass-throughs, spread and duration mechanics, the Harbor Capital government-plus sleeve refactor, a technique decision table, pitfalls, and a checklist. Pair with mortgage-backed securities, investment-grade corporates, and bond duration for the full intermediate fixed-income stack.

Issuer taxonomy and guarantee types

Not all “agency” paper is the same. The label covers GSEs chartered to support housing finance, plus a few federal agencies with explicit Treasury backing. Confusing them leads to wrong spread expectations and tax treatment.

Issuer	Common products	Guarantee	Typical spread vs Treasuries
Ginnie Mae (GNMA)	MBS pass-throughs on FHA/VA loans	Explicit full faith and credit of U.S. government	Tight; often trades like Treasuries plus liquidity premium
Fannie Mae (FNMA)	Debentures, discount notes, MBS	Implicit — conservatorship since 2008; market assumes federal support	+15–60 bps over Treasuries (regime-dependent)
Freddie Mac (FHLMC)	Debentures, discount notes, MBS	Implicit (same as Fannie)	Similar to Fannie; small issuer basis between them
Federal Home Loan Banks (FHLB)	Consolidated obligations, advances	Implicit joint-and-several among member banks	Often inside Fannie/Freddie; deep short-end liquidity
Farm Credit, TVA, others	Sector-specific debentures	Varies; generally high quality, less liquid	Wider; specialist buyer base

Debentures are unsecured bullet or callable bonds — the instrument most investors mean when they say “agency bonds” distinct from MBS. Discount notes are short-term zero-coupon agency paper, competitive with T-bills for cash sleeves. Pass-through MBS are covered in depth in the MBS guide; this article focuses on debentures and the spread decision vs pure government and corporate sleeves.

Why agencies yield more than Treasuries

Agency debentures carry a spread over comparable-maturity Treasuries because they lack the constitutional full-faith-and-credit pledge. Markets compensate for:

Legal ambiguity — Congress could theoretically allow a GSE failure without a bailout; spreads widen when that tail risk feels real.
Lower liquidity — on-the-run Treasuries are the global collateral standard; agency debentures have decent but thinner secondary markets, especially off-the-run issues.
Callable structures — many agency issues are callable at par after a lockout period; investors demand extra yield for negative convexity when rates fall.
Convex supply — GSE issuance rises with mortgage originations; heavy calendar weeks can cheapen new issues briefly.

In calm periods, 5-year Fannie callable debentures might yield 25–40 bps over the 5-year Treasury note. In stress (2008, March 2020, regional-bank scares), spreads can blow out to 100+ bps before recovering as the implicit guarantee narrative reasserts. Agencies typically widen less than IG corporates in credit panics — the behavior Harbor missed when it lumped agencies with corporates instead of government-plus.

Duration and rate risk

Agency debentures obey the same price-yield math as other bonds: modified duration estimates mark-to-market sensitivity. Callable agencies exhibit shortened effective duration when rates fall (issuer calls away high coupons) and extended duration when rates rise (call becomes unlikely). That asymmetry differs from bullet Treasuries and must be modeled, not ignored.

Callable vs bullet agency debentures

Feature	Callable agency	Bullet agency
Yield pickup	Higher coupon/spread for call option sold to issuer	Lower spread; purer duration bet
Rate-down scenario	Likely called at par; reinvest at lower rates	Price appreciates; full duration exposure
Rate-up scenario	Behaves like longer bond; spread may widen	Standard duration loss; hold-to-maturity OK
Best use	Income-oriented sleeves accepting reinvestment risk	Duration targeting; liability matching

Harbor's refactor favored callable Fannie 5s and 7s for 30% of the government-plus bucket — accepting call risk in exchange for 38 bps average spread over Treasuries at entry, with a rule to rotate into bullets if the 5-year Treasury yield drops more than 75 bps in a quarter (call probability rises).

How to buy agency bonds

Channel	Best for	Watch-outs
Brokerage secondary market	Specific Fannie/Freddie CUSIPs; ladder construction	Markups on small lots; verify call schedule and next call date
Agency debenture ETF (e.g. AGZ-style)	Diversified GSE exposure in one ticker	Duration drifts; may hold MBS-like structures — read holdings
Broad aggregate bond fund	Core allocation with implicit agency sleeve	Agency weight is manager-dependent; often mixed with MBS
Institutional new-issue program	Size and spread at auction	Retail rarely accesses; $100k+ minimums typical

Minimum denominations are often $10,000 face — higher than Treasuries at $100 increments via TreasuryDirect. ETFs democratize access but blur the line between debentures and MBS; check the fund's effective duration and sector breakdown before substituting for individual ladders.

Harbor Capital government-plus sleeve refactor

The post-mortem identified three misclassifications:

Binary government bucket — Treasuries-only left spread on the table without adding meaningful credit risk vs a 10% IG sleeve.
MBS confusion — analysts avoided all agency product because prepayment math scared them; debentures do not have CPR risk.
Tax oversight — agency interest is state-tax exempt like Treasuries in many states (verify locally); corporates are fully taxable.

The refactor split the intermediate government sleeve: 50% T-notes ladder, 30% callable Fannie/Freddie debentures (5–7 year weighted), 20% GNMA MBS via a short-duration fund for explicit government guarantee and mortgage-sector diversification. Rebalance trigger: if agency-Treasury spread on 5-year callables compresses below 15 bps, shift 10% into T-notes; if spread exceeds 60 bps, add agencies up to 35% cap. Credit sleeve (IG corporates) unchanged at 15% of total fixed income.

Technique decision table

Approach	Best when	Weak when
Agency debentures (callable)	Want spread over Treasuries; accept call/reinvestment risk; state-tax-sensitive taxable account	Need pure duration; rates falling fast (call away); tiny lot sizes
Agency debentures (bullet)	Spread pickup with clearer duration; hold-to-maturity	Spread too tight vs Treasuries; liquidity before maturity needed
Treasury notes only	Maximum legal clarity; repo collateral; no spread risk	Income target unmet; leaving safe spread on table
IG corporates	Higher yield; sector diversification	Credit panics; need government-adjacent behavior
Agency MBS pass-throughs	Mortgage yield; explicit GNMA government guarantee	Prepayment uncertainty; negative convexity (see MBS guide)
Agency ETF wrapper	Small accounts; daily liquidity	Precision on callables vs bullets; fee drag vs individual bonds

Common pitfalls

Treating Fannie/Freddie like Treasuries — they are not full faith and credit; spreads exist for a reason.
Ignoring call schedules — yield-to-worst matters more than yield-to-maturity on callables.
Confusing debentures with MBS — prepayment risk lives in pass-throughs, not in straight agency debt.
Chasing spread at cycle peaks — wide spreads often mean stress; size the sleeve, do not double down blindly.
Small-lot markups — a 40 bp spread vanishes if you overpay 25 bp on purchase.
Duration mismatch with liabilities — callable effective duration shifts; stress-test +100 bps.
Assuming perpetual implicit guarantee — political risk is low but non-zero; read conservatorship terms.
ETF holdings surprise — “agency” funds may be 80% MBS; check fact sheet duration.

Portfolio checklist

Classify issuer: Ginnie (explicit) vs Fannie/Freddie (implicit) vs FHLB.
Choose debenture vs MBS based on prepayment tolerance.
Compare yield-to-worst on callables vs YTM on bullets.
Benchmark spread vs on-the-run Treasury of similar maturity.
Model effective duration including call probability.
Verify state tax treatment vs corporates in your jurisdiction.
Set spread thresholds for add/trim (e.g. 15–60 bps band).
Cap agency sleeve as % of government bucket (Harbor uses 30–35%).
Hold GNMA separately if explicit guarantee is required.
Document hold-to-maturity vs tradeable agency positions.
Rebalance when duration drifts more than 0.3 years from policy.
Review conservatorship and policy headlines quarterly.

Key takeaways

Agency debentures are government-adjacent, not government-identical — spread compensates guarantee ambiguity and liquidity.
Ginnie is explicit guarantee; Fannie/Freddie are implicit — do not lump them together blindly.
Callables trade extra yield for reinvestment and convexity risk — model yield-to-worst.
Agencies often outperform corporates in credit scares — useful middle sleeve between Treasuries and IG.
Separate debentures from MBS — different risk engines, different guides.