Allocating a 2.4x Oversubscribed Corporate Bond

Start from the constraint: the policy is silent, so the firm is writing the rule while the book is open. That is allowed only if the rule is objective, documented, disclosed to the issuer and applied consistently to everyone.

A workable approach:

• Pro-rata by client category. A uniform scaling of client orders (roughly 41.7% against the total book) is the default that treats like clients alike and is easy to explain to the issuer and to anyone who complains.
• Proprietary order - subordinate or excluded. The firm should not compete with the clients it is distributing to for a scarce allocation. If the issuer is content for the dealer to hold J$150m, that decision must be the issuer's, recorded in writing, and disclosed as a conflict.
• Employee orders - lowest priority and pre-cleared. Personal account dealing policy applies; pre-clearance must have been given before the order was entered, and staff should generally be scaled back harder than clients, not less. If the policy did not contemplate an oversubscribed new issue, allocating to staff at all is questionable.
• Relationship clients - no preference. Fifteen years of business is not a basis for a larger allocation unless a disclosed policy says so in advance and applies to a defined class.

The distribution must be defensible to the issuer, to the regulator and to the scaled-back client.

Reject them, and document why.

• The close is the close. Accepting J$140m of orders after the book closed dilutes every timely subscriber - each existing order would be scaled down further so that a client who missed a published deadline could be accommodated. That is a direct transfer of value from compliant clients to a non-compliant one.
• It creates a precedent problem. Once one late order is accepted, the deadline ceases to exist and every future book is open to negotiation.
• Fair process beats courtesy. The clients should be told promptly, in writing, that the book closed at the stated time, that their orders were not allocated, and that the rejection was recorded. Offering them a place in any subsequent tranche is a legitimate commercial response; back-dating their allocation is not.
• Check the cause. If a firm system failed (an email gateway, a mis-stated deadline on a term sheet), the compliance officer should establish that before the decision is finalised, because a firm error changes the fairness analysis - but it does not change it enough to dilute other clients without their knowledge.

Record the timestamps, the receipt trail, the decision-maker and the notification to each client. 'We squeezed them in' is the answer that turns a small operational question into a distribution-breach finding.

The file should let an outsider reconstruct the decision without interviewing anyone.

• Every order with timestamp, channel, client identity, amount and any conditions attached.
• The allocation policy applied, the version in force, and the fact that it was disclosed to the issuer before allocations were struck; if a policy was created for this issue, say so and say who approved it.
• The scaling calculation, showing the total book, the pro-rata factor and every adjustment made, with the reason for each.
• Conflicts register entries: the proprietary order, the employee orders, the FX relationship, the long-standing pension client.
• Issuer consent for any allocation to the firm's own book, and the issuer's stated register objectives.
• Employee pre-clearance records and personal account dealing approvals with dates.
• Late-order rejections and client notifications.
• Complaint handling, including any response to the pension fund, drafted without disclosing other clients' allocations.

Governance: the allocation should be signed off by someone outside the sales chain - compliance or the head of capital markets - with authority to override a producer. If the only signature is the person who earns the fee, the control is not a control.