Suit Up · Asia's institutional desk on tokenisation

TL;DR

This playbook helps you decide whether a specific tokenised asset is fit for use as collateral in a credit, repo, margin, or liquidity facility, either as the lender accepting it or as the holder offering it. Tokenised collateral is the next-wave financialisation primitive in this stack: when tokenised MMFs (money-market funds), deposits, bonds, and gold become recognised collateral inside those facilities, the cost of capital across the institutional stack drops, and tokenisation crosses from settlement-efficiency story to balance-sheet-capacity story. The operational question (is the asset safe to hold and to liquidate?) is largely answered for narrow use cases, with BUIDL on Aave Horizon and tokenised gilts under the UK DIGIT pilot as the worked examples. The regulatory question (is it CCP-eligible? prime-broker-eligible? Basel SCO60 Group 1a?) is still forming, with Basel SCO60 as the load-bearing prudential text and the targeted review under way as of early 2026. Pick the evaluation lens before the asset; the lender's checklist and the issuer's checklist are not symmetric.

Decision frame

Two reader scenarios. The lender's question: an institution evaluating whether to ACCEPT a tokenised asset as collateral against a credit, repo, margin, or liquidity exposure. The audience is the risk team; the output is yes-with-haircut, yes-with-conditions, or no-with-reasons. The dominant questions are legal-control enforceability in default, valuation reliability under stress, liquidation depth, and capital treatment under the lender's prudential regime.

The issuer's or holder's question: an institution evaluating whether its tokenised asset can BE PROVIDED as collateral, and what design or disclosure changes move it from rejected to accepted, or from accepted-with-thick-haircut to thin-haircut. The audience is the product team or treasury function. The two sides answer the same questions from opposite chairs.

Three framings to avoid. "DeFi proved tokenised collateral works" is wrong because DeFi proved it works in DeFi, where the lending engine is a smart contract and the liquidation engine is automated; institutional collateral runs through a Master Repurchase Agreement or prime-broker margin agreement, with court orders or contractual netting clauses as the liquidation engine, and the supervisory wrapper is real. "If it qualifies for Basel Group 1a, it's collateral-eligible" is wrong because capital treatment and collateral eligibility are correlated but distinct: Group 1a says the bank can hold the asset at viable capital cost, while collateral eligibility says a specific lender, CCP, or prime broker accepts it inside a specific product, which is a separate decision in the venue's collateral schedule. "If it's tokenised, it must be more efficient as collateral" misreads what tokenisation does: it reduces some operational frictions (intraday mobility, T+0 substitution, atomic DvP) and adds others, particularly cross-chain bridge risk and oracle dependency. Both sides have to be priced.

The eight evaluation dimensions

Same dimensions on the lender side and the issuer side. The weightings differ, the questions do not.

Legal recognition of the chain entry as transfer-of-control

In a default scenario, can the lender actually take possession of the collateral. Per Chapter I, Part 3 on legal control, this requires the chain entry to be the legally operative record in the relevant jurisdiction. If the chain entry is a derivative reference to an off-chain register held at a custodian or transfer agent, the lender is taking secured-creditor risk against the off-chain register, not against the on-chain entry. That is a different position from what the asset's marketing usually suggests, and the documentation has to reflect which applies. The US position runs through UCC Article 12's controllable-electronic-record concept; English law through the Property (Digital Assets etc) Bill and preceding case law; Singapore through SFA-recognised electronic securities and Project Guardian's accumulated guidance; Hong Kong through the Personal Property Securities Ordinance and contractual designation, with no codified CER analogue. One tokenised asset can be a clean perfected security interest in one jurisdiction and a contractual claim against an off-chain custodian in another.

Valuation methodology and price discovery

What is the binding price source under the collateral agreement, and what is the fallback if the primary fails. For tokenised MMFs, daily NAV (net asset value) from the fund administrator is canonical, with fallback to the underlying portfolio mark via the primary custodian. For tokenised sovereign or supranational bonds, secondary-market mark requires the underlying market's liquidity to transfer to the wrapper, which it usually does for gilts and Treasuries and rarely does for less-traded sovereign issues. For tokenised gold, the LBMA spot reference is well-established and the wrapper inherits the discipline. For tokenised private credit, secondary-market mark does not exist and valuation runs through the issuer's own NAV process, which is the binding constraint for the asset class. A daily NAV from a regulated fund administrator is institutionally robust; an oracle reading from a single price feed without redundancy is not.

Haircut treatment

The lender's haircut combines three layers: the base regulatory haircut for the asset class under existing frameworks (ECB, Fed, BoE collateral schedules being the canonical references for sovereign bonds), an additional tokenisation premium for structural-novelty risk, and a liquidity haircut for the depth of the liquidation path. As tokenisation matures and structural-novelty risk gets tested rather than theoretical, the tokenisation premium compresses; the premium is meaningful in 2026 but trending down. Specific numbers are not consistently published; reason from the structure rather than from any quoted figure.

Liquidation path

In default, can the lender sell into a market of meaningful depth under stress. For BUIDL, the path runs through Securitize Markets (a registered alternative trading system) with primary redemption to BlackRock as back-stop; liquidity is meaningful but not bulge-bracket-deep, and large blocks remain price-impact-sensitive. For tokenised UK gilts under DIGIT, the path is intermediated through the conventional gilt market with the wrapper alongside the conventional registered holding. For tokenised gold, the path is the wrapper issuer's redemption rail plus bilateral OTC. For tokenised private credit, the path is bilateral OTC in practice, which is the binding constraint for the asset class as collateral.

Custody arrangement

Who holds the collateral on the lender's behalf, and is the custody segregated, bankruptcy-remote, and regulator-supervised. The chain entry being the operative record does not relieve the custody question; the custodian still has to be the right entity under the right licence with the right operational discipline. See Evaluating custody providers. The lender-side custodian is often a different entity from the issuer-side custodian, and the operational handshake (whitelist consistency, transfer-agent integration, settlement coordination) is where integration cost concentrates.

Cross-chain considerations

If the collateral lives on chain A and the loan on chain B, the bridge IS the credit exposure, and bridge security has not been the strongest part of the public-chain ecosystem historically. Most institutional programmes today insist on same-chain pairing; cross-chain collateral introduces bridge risk prime-broker risk teams have not yet calibrated. BUIDL's nine-ledger availability (per BUIDL reference architecture) is a distribution feature but a collateral-pairing constraint: the lender locks chain identity at agreement time and stays there.

Regulatory recognition

Three perimeters. CCP (central counterparty) eligibility is the highest hurdle; CCP collateral schedules have not generally admitted tokenised assets as of 2026, and where they have, it has been in pilot mode rather than standing eligibility. Prime-broker eligibility is the more progressed frontier; major prime brokers will accept BUIDL or comparable instruments under bilateral negotiation, schedules being a function of each prime broker's balance-sheet discipline rather than a published industry standard. Capital treatment under Basel SCO60 sits beneath both: Group 1a is the gating condition for the lender to hold the asset at viable capital cost, and the SCO60 targeted review endorsed by GHOS in March 2026 is the principal Basel-side process to track. See Capital treatment of tokenised assets for the prudential decomposition.

Operational integration

How does the lender's risk system actually ingest on-chain price and position data. Most lender risk systems were built for off-chain feeds; integrating an on-chain oracle, whitelist read, or mark-to-market signal into a Murex or Calypso instance is non-trivial engineering. Expect 3-6 months per asset class for first-time integrations, with subsequent asset classes amortising onto the same plumbing. The integration cost is an under-discussed reason institutional collateral acceptance moves slowly: legal and prudential work clears, then risk-system build is the binding gate.

Asset-class scorecards

Collateral readiness varies sharply by asset class. The differences track the underlying asset's existing collateral discipline and the maturity of the tokenised wrapper.

Tokenised MMFs sit at highest readiness. BUIDL accepted on Aave Horizon alongside Janus Henderson JAAA and JTRSY plus Securitize-issued credit products is the worked example, with bilateral prime-broker acceptance emerging through 2025-2026. Daily NAV from a regulated fund administrator, regulated-asset-manager issuer, transparent reserve composition. Aave Horizon's April 2025 launch plus the November 2025 Binance acceptance of BUIDL as institutional collateral mark the crossing from "MMFs are collateral-shaped" to "MMFs are accepted collateral inside named venues."

Tokenised deposits have limited collateral application today, despite a growing role on the cash leg of DvP transactions. Deposits are bank money inside the issuing bank's ledger, which makes them operationally simpler as the cash leg of a DvP trade than as collateral against external credit; the bank already gives the depositor a claim against its balance sheet, and using that claim as collateral against external credit introduces a circularity the treasury team would rather avoid. Cross-bank pilots exist (the DBS-Kinexys interoperability framework bilaterally, HKMA EnsembleTX for regulator-coordinated wholesale settlement) but standing acceptance across rails is early.

Tokenised bonds have strong readiness when the underlying is sovereign or supranational, because secondary-market depth transfers to the wrapper. UK DIGIT pilot tokenised gilts (per HSBC Orion + DIGIT) is the G7 sovereign reference; the HKSAR multi-currency green bond programme on Orion is the APAC analogue. The Project Ensemble repo workstream is exploring tokenised bond collateral on the wholesale side. For non-sovereign issuance, readiness drops sharply because secondary-market depth is the binding constraint.

Tokenised gold sits at mature readiness. PAXG and HSBC's tokenised gold programme are accepted in bilateral arrangements between sophisticated counterparties, the LBMA spot reference is canonical, and underlying custody is institutional-grade. The asset class is small but the per-unit collateral discipline is among the cleanest in the tokenised-collateral universe.

Tokenised private credit has moderate readiness on DeFi rails and limited acceptance in CCP and prime-broker programmes. Centrifuge V3 plus Aave Horizon for JAAA and JTRSY is the worked example, with Sky's Grove allocator deploying USD 250m of JAAA on Avalanche as the canonical DeFi-native case. CCP and prime-broker acceptance lags because the underlying credit is illiquid and valuation is less canonical, which compounds in stress.

Tokenised equities sit at low readiness, because the legal structure for an equity that is also an on-chain controllable electronic record is not in place in most jurisdictions. Wrapped equity products (Robinhood EU stock-tokens, Ondo Global Markets) are derivatives referencing the underlying through a contractual claim against the wrapping party, and they inherit derivative-collateral treatment. Not where institutional tokenised collateral is being built today.

Worked example: APAC GSIB prime-brokerage desk evaluating BUIDL as repo collateral

An APAC GSIB's prime-brokerage desk evaluates whether to accept BUIDL as repo collateral against HKD or CNH funding for an asset-manager client. The desk runs the eight dimensions.

Legal recognition. BUIDL is a 3(c)(7) wrapper with Securitize as SEC-registered transfer agent and BNY as custodian of the underlying T-bill, repo, and cash sleeve; the on-chain entry is the operative record for the transfer-agent function under UCC Article 12 in the relevant US states. For HK-side perfection, the bank takes a security interest under HK common-law trust principles plus contractual designation, with US-side perfection in parallel. Legal-opinion budget is real but the architecture is workable.

Valuation. Daily NAV from BlackRock as fund administrator with secondary-market mark via Securitize Markets as cross-check; primary redemption to BlackRock as fallback. The lender's risk system picks which feed is binding.

Haircut. Start at the conventional haircut for short-dated US Treasury paper, layer a tokenisation premium for structural-novelty risk, add a liquidity haircut reflecting Securitize Markets depth. Combined haircut is meaningfully wider than for conventional T-bill collateral and has room to compress as the relationship matures. Specific numbers are bilateral and not published.

Liquidation. Securitize Markets ATS as secondary path, primary redemption to BlackRock as deeper back-stop. Block-trade price impact is the binding constraint for large positions; the desk sizes maximum acceptable position accordingly.

Custody. Anchorage as the on-chain custodian for the prime-broker arrangement, with the client's BUIDL transferred to the bank's whitelisted address inside Anchorage's segregated infrastructure for the duration of the repo. The Securitize-Anchorage transfer-agent handshake is the load-bearing operational integration.

Cross-chain. BUIDL is multi-chain across nine ledgers; the desk locks the loan and the collateral to the same chain identity at agreement time. The bank does not accept cross-chain collateral pairing in the first generation of the programme.

Regulatory. Basel SCO60 Group 1a treatment for the bank holding the BUIDL position. HK acceptability is workable as bilateral OTC under the bank's existing licensed activities. CCP eligibility is not in scope; the repo runs bilaterally against the asset-manager client.

Operational. A 3-month build to ingest BUIDL price feed from BlackRock, on-chain position from Anchorage, and secondary-mark feed from Securitize Markets. One-off; subsequent tokenised-MMF acceptance amortises onto the same plumbing.

The recommendation: accept BUIDL as repo collateral under bilateral arrangement with the named haircut, chain-identity-lock, and a 12-month review for expansion. The gate covers two questions: whether operational discipline holds in production, and whether the prime-broker franchise wants to extend the same architecture to FOBXX, OUSG, and JAAA on the same plumbing.

Red flags

Cross-chain collateral with implicit bridge risk. If collateral on chain A is being lent against on chain B with a bridge in between, the bridge IS the credit exposure. Most institutional-grade programmes lock chain identity at agreement time for this reason.
Oracle dependency without a fallback price source. A single oracle feed without redundancy is a single point of failure for the binding price under the collateral agreement; the documentation should specify the fallback hierarchy explicitly.
Opaque custody arrangement on the on-chain leg. The chain entry can be the operative record under the relevant property law and the custody arrangement can still be the binding constraint. Licence, segregation model, insurance posture, and bankruptcy-remoteness are independent dimensions, even when the on-chain mechanics are clean.
Asset class without a secondary market for liquidation. If the only path to liquidate in default is bilateral OTC at the issuer's discretion, the collateral is not really collateral in a stress scenario; it is a contractual claim against the issuer's good behaviour.
Jurisdictional non-recognition of on-chain control. The collateral can be perfected on-chain in the issuer's jurisdiction and still be unenforceable in the court that hears the borrower's insolvency. Cross-border arrangements need parallel legal opinions, and the cross-jurisdictional gap from the legal-control chapter is the operational expression.
DeFi-pattern collateral acceptance imported into institutional context without the supervisory wrapper. The DeFi liquidation engine is automated and hard-coded; the institutional liquidation engine is a court order or a contractual netting clause. Importing the DeFi pattern (low haircut, fast turnover, high composability) without the supervisory and contractual scaffolding creates a structural mismatch that surfaces in stress.

[Suit Up]

Evaluating tokenised assets as collateral