The Books They Manage

IntroductionThe posting, and the question

The Fund posted an Assistant Corporate Bond Portfolio Manager role on 27 October 2026, with applications due 1 December 2026.

The posted salary is $179,150 to $229,150 a year, full time and permanent, at 199 Church Street, New York. The posting says the Fund manages more than $22 billion and that the corporate bond exposure is more than $9 billion. The official civil-service title is Special Investment Officer. Candidates send a résumé and cover letter to the Fund’s recruiting address.

The question this chapter asks is the one the role exists to answer: what is in the books the role touches, and why did they move the way they did, according to the Fund’s own filings?

Section 1 · 31 December 2025What is in the books

A bond book run against claims that are paid over decades.

The Workers’ Compensation Fund reported $23.4 billion of cash and invested assets at the end of 2025. Bonds were $19.47 billion of that, or 83.2 percent; common stocks were $2.42 billion (10.4 percent); cash and short-term investments were $945 million (4.0 percent). Corporate bonds were $9.52 billion, 49 percent of the bond book, which agrees with the posting’s “more than $9 billion.”

From 2025 the report uses new bond categories under the revised statutory definition of a bond, and says itself that 2024 figures “are not directly comparable”; the table is not compared with earlier years for that reason. By rating, 91.1 percent of the bonds carry the top NAIC designation and 4.5 percent the second; $851.7 million, 4.4 percent, is below investment grade. These shares are of the bond book; the scanned page prints its own percentages against total admitted assets of $25.0 billion, such as 70.9 percent for the top designation. Those designation amounts come from a scanned page and add to $5.2 million more than the balance-sheet bond figure; the report does not explain the difference. The largest single exposures are Freddie Mac ($1,042.5 million) and Fannie Mae ($765.2 million). The Fund also had $509.7 million of securities on loan, supported by $521.9 million of collateral; its lending agent is JPMorgan Chase.

Sources: 2025 annual report, balance sheet (p. 4), Note 3 (p. 24 for the bond table, p. 25 for maturities, pp. 26–27 for securities lending) and the investment interrogatories (p. 46). The last two table columns are the report’s own; the percentages are the author’s arithmetic.

Section 2 · Policy and lawThe rules the books run under

A policy that has moved since, and a rule from the law that has not.

The Fund’s investment policy statement, in the version amended on 18 November 2020, sets the frame for the main Workers’ Compensation Fund. Investment-grade fixed income must be between 78 and 100 percent of market value, with a target of 85. The other targets are 4 percent high yield, 5 percent public equities (maximum 8), 4 percent alternatives and 2 percent cash. The fixed income portfolio’s weighted average rating must be at least mid-single A (A2), and its duration target is 8 years, plus or minus 1. The policy asks for enough U.S. government securities to cover 75 percent of liabilities due within a year, allows securities lending under the Workers’ Compensation Law, and allows derivatives only with surplus funds. The policy statement itself contains no issuer or sector limits; any such limits would sit in manager guidelines or other documents not read here.

One rule comes from the law, not the policy. The Workers’ Compensation Law makes the Fund discount all its claim reserves at a fixed 5 percent (2025 report, p. 9). The reserves assume money earns 5 percent; whether the bonds earn that is a separate matter, taken up in Section 4.

Section 3 · Fair value against bookWhat the numbers say, and what they do not

The gap sits at the long end, and the Fund says why.

At the end of 2025 the bond book was $1.23 billion below its book value, about 6.3 percent. U.S. Government obligations, whose credit risk is small, were about 21 percent below book; municipal bonds about 11 percent; corporate bonds 2.5 percent; asset-backed securities 2.3 percent. A reader who looks for the losses in the corporate bonds would find less than a quarter of them there.

The maturity table shows where the gap sits. Bonds due after ten years were $1.17 billion below book, 95 percent of the whole gap. Bonds due in five to ten years were $15.4 million above book, and bonds due within a year were within 1 percent of it. The Fund’s own note says the same in words: the securities in a loss position “are primarily investment grade securities with extended maturity dates, which have been primarily impacted by a change in interest rates after the purchase date” (p. 29). The table does not split the corporate gap into its rate and credit parts, so how much of that is credit is not measured here.

These are differences between fair value and book value on a book run against long-dated claims. They are not realized losses. Bonds sold below book are realized losses, and those are the subject of Section 4.

Section 4 · 2019 to 2025Follow the money

Why would a fund sell bonds below cost? The filings rule some reasons out and fit others.

Every annual report shows its own year and the year before, so the 2020 to 2025 reports give seven years of the same lines. The table puts them side by side: realized gains and losses on stocks and on bonds, the stock share of cash and invested assets, bond investment income divided by year-end bond book value, and proceeds from investments sold, matured or repaid divided by bond book value.

Stocks gave gains almost every year

They produced realized gains in six of the seven years and a net $1.01 billion over the seven; only 2022, when both markets fell, shows a loss. The regulator’s reports show the same habit earlier, without splitting it by asset: net realized gains of $1.14 billion over 2010 to 2014 and $957 million over 2015 to 2019. The stock share sat between 8.0 and 10.4 percent of cash and invested assets in every year, above the 2020 policy’s 5 percent target, so selling stocks at a gain may be one way of trimming that excess, although the share did not fall over the period. That is an inference, not something the filings state.

Bond losses are a rate story, not a decade story

Bonds were sold at net gains of $446.1 million over 2019 to 2021, while rates were low, and at net losses of $174.5 million over 2022 to 2025, after they rose. The losses have shrunk each year since 2023: −85.3, −43.4 and −15.0 million.

Nothing in the annual figures forced a sale

Operations brought in cash in every year, and the Fund bought more investments than it sold or saw mature in every year, by $203.7 million at the least. The Fund pays no federal or state income tax, so the usual tax reason for booking a loss does not apply. The 2020 to 2024 reports say no other-than-temporary impairments were recorded; the 2025 report says none were recorded for asset-backed securities and changes its general wording.

The Fund’s own wording on selling at a loss

Each report from 2020 to 2025 carries the same sentence: “securities may be sold at a loss and repurchased within 30 days of the sale date to enhance WCF’s yield on its investment portfolio.” “May be” describes what the Fund allows, not what it did or how often, and the disclosure that follows covers only lower-rated securities. The figures fit that reason: as the bond losses were taken, bond income divided by bond book rose from 2.68 percent in 2022 to 3.67 percent in 2025. Rising rates alone would also lift that ratio, so this is a fit, not a test. Over the same years turnover fell by about half, from 42 percent of the bond book in 2019 to about 19 percent from 2022, so most bonds below cost were held, not sold.

The five percent gap is the long-running part

The reserves assume 5 percent. The regulator’s 2014 report says the Fund’s actual investment yield “has not exceeded five percent during the examination period,” and lists 4.1, 3.8, 3.4, 3.0 and 2.8 percent for 2010 to 2014. The rougher ratio in the table stays between 2.63 and 3.67 percent through 2025. Reading the two together, gains on stocks look like one of the ways the Fund’s results have kept pace with an assumption its bonds did not meet. That is an inference from the filings, not something they state.

Section 5 · A proposalWhat a daily monitor would watch

The author’s proposals, not the Fund’s plans; the monitor does not exist yet.

The 2020 policy benchmarks the main fund’s investment-grade bonds against a “NYSIF Custom Benchmark” whose make-up it does not give; ICE BofA indices appear only for cash and for two smaller funds. A daily note could still track the Treasury curve (especially the 7 to 10 year point and the long end), the option-adjusted spread on investment-grade corporates by rating (A and BBB), and the change in each since the prior day, all from public data.

It could also follow the questions this chapter raises: how closely the bonds’ cash flows track expected claim payments, which is what the duration target is for; how the book’s yield compares with the 5 percent the reserves assume; and, for any bond sold below book, the yield given up against the yield bought. These are questions the role would study, not a recommendation to hedge. The first test of the monitor is whether it reproduces a known past move correctly; the second is whether it tells a person something the morning market summary does not.

LimitsWhat this does not show

• Which bonds were sold, when and why. The list of bonds sold, with each sale’s gain or loss, is in the Fund’s statutory annual statement filed with the Department of Financial Services, not in the annual report, and was not read.
• The Fund’s holdings after 31 December 2025, and the policy now in force.
• How much of any loss is credit rather than interest rates, bond by bond.
• That the author meets the posting’s experience requirement. The author’s background is corporate credit and securities finance, not a bond trading book; the application says so.

UpdatesWhat changed after the first version