Extended analysis · engagement previews · RFP 902732
Three analyses the engagement will finish — started from public data alone.
The same 31,129-row reconstruction behind the recast engine supports analyses that usually wait for custodial data. Below are three, run as working first passes — a preview of week-one of the engagement, not its conclusion.
Working analyses — read accordingly. First-pass results from the County's public monthly reports only. Exit values are estimated at the last reported month-end mark (actual execution prices arrive with custodial data); classifications are heuristic and individually spot-verified against the County's own activity pages where noted. These previews have not yet been through our Lead Consultant's QA/QC. The engagement confirms, corrects, and completes them.
1 · The reconstructed trade log — what the pool actually did, 2020–2026
Tracking every position month-to-month across 74 reports reveals every early exit — each bond that left the book before its maturity date. That is a trade history nobody publishes, reconstructed from documents everybody can read.
1,681
positions tracked over six years — 898 ran to maturity, 401 still held
382
early exits (sold or called before maturity) — $5.7B of par
≈ $0
net estimated realized P&L across all 382 exits (+$3.2M, and +$3.0M of it is a single 2020 trade)
−$8.7M
total estimated realized losses in six years — 0.05% of exited par; no fire sales
AThe call wave has happened before — and it is happening now. In 2020–21, ~$1.5B of agencies (average coupon 1.62%, above-market once rates collapsed) were called away, and the pool's book yield fell from 1.92% to 0.82% in two years. The same mechanic is now running in reverse-era: ~$2.35B of call-pattern exits in 2024–25 at ~5% coupons, including a $50M FHLB 5.19% in the Dec–Jan window and a $50M FHLB 4.50% in February 2026. The five largest were each verified by CUSIP in the County's own activity/redemption pages. This is the yield-durability question made historical fact — twice.
BLoss discipline is real — and pragmatic, not absolute. Of $5.7B in early exits, exits marked below book total an estimated −$8.7M over six years — five basis points of the par exited. Most of it is recent and deliberate: 2024–25 clearing of low-coupon 2020–21-era paper at small discounts, recycling into today's yields. The pool demonstrably does not dump bonds into weakness — and it demonstrably will take a small, sensible loss to improve the book. Both halves matter for how an Ethical Investment Policy divestment would actually be executed.
CThe single best trade in the file: May 2020 — $50M of a 1.50% Treasury maturing 2024, sold at 105.29 at the bottom of the rate cycle, an estimated +$3.0M realized gain. Rich duration sold at the top. Credit where due.
Maps to RFP: performance & management-practices evaluationrisk reviewEIP implementation feasibility
2 · The §27136 withdrawal gauge — who pays when money leaves at book value
Participants withdraw at book value while the pool marks at market. Whenever NAV is below 100, a withdrawal quietly transfers the difference to those who remain — mostly the County itself and the captive districts. The reconstruction prices that transfer for every month since 2020.
36 of 75
months in which the pool's NAV sat below 100 — the transfer window
$4.14M
value transferred to remaining participants per $100M withdrawn at the September 2022 trough (NAV 95.857)
$2.3–2.4M
per $100M through most of 2022–23 — the exposure persisted for two years, not one bad month
−$0.70M
today (NAV 100.7): a book-value withdrawal now leaves value behind — the gauge runs both ways
Method: transfer per $100M = (100 − NAV) × $1M, monthly, from the 75-month NAV series. This extends the run-risk simulator's finding — the structural protection is that the discretionary money is the County's own — by pricing the fairness exposure the book-value rule creates for the captive districts during underwater periods, and the symmetrical giveaway during premium periods.
Maps to RFP: governance & oversight (Gov. Code §27130–37)liquidity / run-risk
3 · The reinvestment runway — what 4.18% becomes, and where the real risk hides
The maturity schedule and each maturing cohort's locked-in yield are public. Rolling them forward prices the question the Board should ask of any yield figure: is it durable?
| Maturity window | Par rolling off | Yield rolling off | Read |
|---|
| 2026 H1–H2 | $1,074M | 3.41–3.68% | below today's market — reinvestment lifts book yield |
| 2027 | $1,454M | 4.14–4.27% | roughly market — neutral |
| 2028–2029 | $4,818M | 4.44–4.57% | the high-coupon towers — where call risk concentrates |
| 2030–2032 | $1,945M | 4.04–4.48% | incl. the seven >5yr policy-basket holdings |
AOn stated maturities alone, the yield is durable. Reinvesting everything that matures through 2027 (~$2.5B) at today's ~4.05% two-year rate moves the coupon book's par-weighted yield from 4.26% to 4.29% — because the near-term roll-off is low-yielding 2023-vintage paper. Even at −50bp the book holds ~4.15%; it takes −100bp to pull it to ~4.01% by end-2027.
BThe durability risk is the calls, not the calendar. $1.63B of agencies trade above par — the 2028–29 high-coupon towers — and the trade log above shows the issuers already exercising: ~$2.35B of call-pattern exits since 2024. A call replaces a 5% coupon with a ~4% one regardless of the maturity schedule. Public data carries no call schedules, so this page can size the exposure but not the timing — loading call schedules and re-running the book to yield-to-worst is a first-week engagement task.
Maps to RFP: performance evaluationrisk & scenario analysisreporting practices
Why these three previews are here.
Each one answers a question the RFP poses — how has the portfolio actually been managed, is the governance structure fair to captive participants, is the yield durable — and each was produced from the County's public record alone, by the same reconstruction that reproduces the County's published totals to the dollar. The engagement does not start these analyses; it finishes them, against custodial and transaction data, under our Lead Consultant's QA/QC.
That is the working model this team is proposing: the analysis arrives early, the verification is institutional, and the Board reads findings — not promises.