I believe a bank should approve the transition plan at the same time as it approves a material reduction in connected commercial exposure. A new limit without a sequence of customer decisions is an incomplete management instruction. It can leave relationship teams trying to satisfy a portfolio requirement through hurried individual conversations, with consequences the board never examined. The credit judgement may be sound while the implementation is unnecessarily damaging. A chief executive is responsible for both parts of that decision.
Connected exposure is easy to discuss in the abstract and difficult to reduce in practice. The loans may have different maturities, security arrangements and repayment sources. Some customers can refinance readily; others require time or changes to their own plans. A bank cannot resolve those differences by announcing that a concentration percentage is too high. It needs to know which commitments it can change, which obligations remain and how its actions will affect the economic dependency it is trying to reduce. Withdrawal can itself change the situation.
Find the common repayment dependency
Our review began with a question about a regional construction ecosystem. The borrowers appeared under several sector codes: materials, transport, property and business services. Their legal identities were distinct. Yet a delay in the same development cycle could affect cash receipts across several of them. Relationship managers understood parts of that connection, but those observations had not been assembled into a portfolio view. No single lending file contained the whole story, and asking each team to confirm its own file did not reveal the aggregate.
We traced the major sources of repayment and the events needed to generate them. That required care. A shared customer or geography does not automatically make two exposures equivalent. We looked at contract terms, alternative buyers, liquidity and the ability to withstand a delay. Some apparent connections were weak; others were more consequential than the original reporting suggested. The aim was not to label everything as connected. It was to identify a set of dependencies that could materially affect several borrowers at the same time.
I asked for the uncertainties to remain visible. Where a relationship manager believed a borrower could replace lost business, we recorded the evidence and the time required. Where information was missing, we distinguished that from evidence of deterioration. This prevented an incomplete file from being treated automatically as a weak borrower while still recognising that uncertainty has a cost. It also made disagreement more useful. Credit and commercial colleagues could debate a specific assumption instead of exchanging broad claims about caution or customer knowledge.
The resulting C$1.35 billion exposure was a starting position for a management choice. It did not mean every facility was impaired or that the bank should abandon the region. I was careful about that distinction in both board and internal communication. A concentration finding can become a reputational judgement about customers if leaders use imprecise language. The bank's need to adjust its balance sheet is not the same as a conclusion that each customer has become unsuitable. Our explanation had to reflect the actual reason for the change.
A limit needs a customer sequence
We considered a rapid portfolio sale and rejected it after examining pricing, customer disruption and evidence a buyer would require. We also rejected a passive approach that depended on ordinary renewals gradually solving the issue. Instead, we proposed a thirty-month sequence using scheduled maturities, selective refinancing and tighter boundaries around additional exposure. That choice carried risk: concentration would remain elevated for part of the period, and some decisions would have to be revisited if conditions deteriorated. We made that residual exposure explicit.
Each material relationship needed a plan. The plan identified the next contractual decision, the bank's intended position and evidence that could change it. Relationship managers received support for customer conversations before formal renewal discussions began. We asked them to explain the bank's position without making promises about another lender's willingness to provide funding. Where a customer's independent repayment capacity justified continued support, we documented the distinction. Where the common dependency remained material, we avoided using familiarity as a reason to postpone the difficult conversation.
Commercial colleagues reasonably asked what business would replace the income we were giving up. I did not want the answer to be an unrelated volume push that created a new concentration elsewhere. We set a more modest growth expectation and tracked the forgone origination income. That was a chief executive decision, not an instruction for relationship managers to absorb an impossible combination of lower exposure and unchanged revenue. If management approves restraint but leaves every commercial measure untouched, it should expect exceptions and inconsistent behaviour to follow.
The sequence also required operating capacity. Customer notices, documentation changes and monitoring work take time. We made those tasks part of the programme rather than treating them as administrative details after the credit decision. A delayed notice can turn a planned transition into a rushed negotiation. A missing record can leave the next relationship manager unaware of what the bank has already explained. I learnt from earlier roles that implementation quality often determines whether a prudent decision is experienced as fair and predictable or arbitrary and confused.
Keep the exceptions in the arithmetic
Every transition produces requests for exceptions. Some are sensible. A temporary extension may allow an orderly refinancing; a facility with genuinely independent cash generation may deserve different treatment. The problem arises when exceptions are considered individually and disappear from the overall reduction plan. We maintained a bridge from the original exposure to the current position, showing repayments, new drawings, renewals and approved additions. That made it possible for the board to see whether the direction remained credible after allowing for the decisions management had actually made.
I wanted commercial and risk executives to present that bridge together. Separate presentations can create the appearance of agreement while leaving different assumptions unexplored. Joint ownership did not require identical views. It required each executive to understand the other's evidence and to bring unresolved differences into the decision. Where I approved an exception, the rationale and effect stayed in the record. A chief executive should not be able to support a portfolio policy in public and quietly undermine it through a series of private accommodations.
By June 2026, the connected exposure had fallen to C$980 million without a forced portfolio sale. I regard the process that produced that reduction as important as the number itself. We now ask about shared repayment dependencies earlier and keep the customer sequence visible when changing appetite. There are still judgements to make about the remaining portfolio, and the original plan cannot be assumed suitable indefinitely. Changes in customer trading, refinancing availability or the construction cycle may require a different pace and a fresh board discussion.
My advice to another bank leader is to request three views before approving a large concentration reduction: the economic dependency, the contractual sequence and the commercial consequence. If one is missing, the proposal is not yet ready to become an instruction. A lower exposure figure is meaningful only when management can explain how it will be achieved and what the bank is accepting along the way. The customer plan is part of the credit decision because customers, colleagues and shareholders ultimately live with its implementation, not its wording.