Lending a share does not remove the price exposure
In securities lending, an owner temporarily transfers securities to a borrower that must return equivalent securities, usually against collateral. An agent can arrange loans for an institutional owner, while a broker may borrow as principal from its customer. Borrowers use securities for short-sale delivery, market-making and other settlement needs. The New York Fed’s reference guide describes these arrangements and the distinction between borrowing an asset and borrowing cash. [1]
The owner’s economic exposure generally remains: if the share price falls, an extra lending fee does not cancel the loss. What changes is the legal and operational route through which the owner gets securities back, receives equivalent distributions and exercises rights. Calling the transaction income on an existing holding is accurate only if those additional exposures remain visible.
The chain of counterparties matters
An institutional program might connect a pension fund, its lending agent and a borrowing dealer. A retail program may instead make the brokerage firm the customer’s direct borrower, even if that firm lends onward. The borrower shown on the owner’s agreement determines the immediate return obligation; the identity of an ultimate short seller does not automatically replace that contractual relationship.
Fidelity’s current fully paid lending explanation illustrates one specific design: Fidelity borrows as principal, maintains collateral at a third-party custodial bank, adjusts it daily and permits sale or recall. It says loaned shares lack SIPC coverage and describes collateral as at least 100% of loan value. These are that provider’s stated terms, not a claim that every program uses the same percentage or protection. [2]
Analysis: a long chain is not inherently unsafe, but it can complicate the answer to who owes what after a failure. Separate obligations connect owner to borrower, borrower to its customer, and collateral provider to custodian. The fact that a security has been passed onward does not create a second vote for the original owner or make every participant directly liable to that owner.
Fee income and cash-collateral rebates
When collateral is noncash, the borrower commonly pays an explicit lending fee. In cash-collateral arrangements, the lender or its agent may invest the cash and pay the borrower a rebate, retaining a spread after expenses and revenue sharing. Scarce securities can command different economics from abundant ones. The reference guide describes both collateral forms and the reinvestment exposure that cash introduces. [1]
Consider a hypothetical $10 million loan with noncash collateral, a 0.60% annual fee and a 30-day period using an actual/360 convention. Gross fee income is $5,000. If the agent contract retains 20%, the owner receives $4,000 before other expenses or taxes. If the loan ends after ten days, the same assumptions produce only $1,333.33 for the owner. An annualized quoted rate is not a promise of a full year’s utilization.
In a separate invented cash-collateral example, $10.2 million earns 4.50% while the borrower receives a 4.00% rebate. Over 30 days on the same convention, gross investment income is $38,250, the rebate is $34,000 and the spread is $4,250 before agent fees and losses. These examples describe alternative structures; combining their income would double-count two different assumptions.
Collateral protects replacement, not investment performance
Suppose $10 million of loaned shares is backed by $10.2 million cash. The shares then rise 3% to $10.3 million before additional collateral arrives. If the borrower fails at that instant, the original cash is $100,000 short of the new replacement cost, before closeout expenses. The illustration assumes no collateral interest and immediate replacement; real settlement timing and contract rights can alter the outcome.
Now isolate another risk. If the $10.2 million cash-collateral investment loses 1%, it falls by $102,000. Under the earlier 0.50% gross spread assumption, that loss equals 24 of the illustrated 30-day spread amounts: $102,000 divided by $4,250. The borrower might return the securities in full and still expect its cash back. Reinvestment loss is therefore different from borrower default.
This distinction explains why collateral quality and the loaned security’s quality cannot be collapsed into a single judgment. Noncash collateral avoids that particular cash-investment structure but can itself fall in value, be difficult to sell or move adversely relative to the asset being replaced. Collateral reduces an exposure; it does not automatically extinguish it.
What the U.S. customer-protection rule actually adds
For covered broker-dealer borrowing of fully paid or excess-margin customer securities, Rule 15c3-3 requires a written agreement with terms governing the loan, qualifying collateral that fully secures it and marking the loan to market. FINRA’s published rule text and interpretations also require a prominent warning that SIPA may not protect the lender and collateral may be the only source of satisfaction if the broker fails to return the securities. That framework does not insure against a decline in the investment’s market price. [3]
The important analytical boundary is between a rule’s minimum protections and a program’s additional promises. A stated indemnity, for example, is only as broad as its covered events and the party’s ability to perform. It cannot be assumed to cover collateral-reinvestment losses, lost votes, every tax effect or every operational delay. Those are agreement-specific questions rather than generic attributes of securities lending.
Recalls, voting and equivalent payments
Fidelity’s disclosure says customers give up voting rights while shares are on loan over a proxy record date and can recall in advance of that date. It also explains that distributions on borrowed shares can be paid as cash in lieu, with potentially different tax treatment. A payment with the same dollar amount need not have the same legal or tax character as an issuer-paid dividend. [2]
Analysis: a recall creates a delivery obligation, not instantaneous physical possession. The borrower may need another lending source or may close the position. A rush of recalls in a scarce security can therefore affect borrowing costs and trading behavior. Whether a particular recall meets a voting or sale deadline depends on the contract, notice timing and successful delivery; a general recall right alone does not prove that every deadline will be met.
FINRA’s December 2023 SogoTrade settlement provides a historical example of the disclosure problem. It addressed misleading statements about customers’ compensation and deficient supervision of a fully paid lending program. SogoTrade accepted FINRA’s findings without admitting or denying them. The settlement concerns that specific case, not every broker offering the product. [4]
The incremental economics are the right comparison
The meaningful comparison is the same portfolio with and without lending, including actual days on loan, realized fee sharing, collateral costs and any changed treatment of distributions. In the hypothetical noncash example, $4,000 earned on $10 million is 0.04% for the 30-day period. A 2% fall in the securities would be $200,000. Lending income would partly offset that decline, not transform the holding into a protected instrument.
High borrowing fees may reflect scarce supply, a corporate event or concentrated demand; they do not establish that the borrower’s negative view of the company is correct. Conversely, modest fees do not prove that a program has no operational risk. Evidence about utilization, collateral performance, successful recalls, counterparty exposure and actual net revenue clarifies what the lending activity added. The loan is an extra layer on the original investment, with its own cash flows and obligations.
Sources
- New York Fed Staff Report 740, Reference Guide to U.S. Repo and Securities Lending Markets, September 2015, revised December 2015Official source · PDFBack to text: ↑1↑2
- Fidelity, Fully paid lending program explanation and risks; provider disclosure checked October 4, 2026SourceBack to text: ↑1↑2
- FINRA, SEA Rule 15c3-3 and Related Interpretations, paragraph (b)(3); checked October 4, 2026SourceBack to text: ↑
- FINRA, SogoTrade acceptance, waiver and consent, December 2023; historical disclosure and supervision findingsSource · PDFBack to text: ↑