The Losses That Didn't Happen

Start with who is listening. A rating is written about one person and read by everyone: it is the loudest thing a manager ever says about what judgment is for. So the question is not only what this year was worth. It is what the team will do next year once it knows what was counted.

A counts what can be seen, which is fair to the files that have resolved and blind to the ones that cannot have. C counts the best story, the one with a grateful client as its witness — and it was genuinely good work. Both teach the same lesson: prefer decisions that produce a witness soon. An approval produces one within the year. A decline never does; from the outside it cannot be told apart from timidity, and a team that learns it goes unrewarded will make fewer of them, which is how a lender's losses begin a few years before anyone sees them. D sounds rigorous and says the loudest thing of all: that on a year like this one, “meets expectations” is the ceiling until the market has marked the work.

The senior banker in the reviewer's chair chooses B. Ask for the record — every decline and every structure held, with the reasons written at the time, not reconstructed for the meeting — and rate the quality of the reasoning now. Then keep the record, and reread it against what happened, next year and the year after. Most entries will resolve dull, which is useful: it calibrates. Some will show the decline was wrong, and that belongs in the record too; a record that is never wrong is not a record. The occasional entry will resolve into a loss at another lender, and that is the only way the discipline of saying no is ever measured.

If there is no written record, that is the review. Not a criticism of the calls — a request that next year's exist on paper, because a manager can reward only what can be seen, and a banker's best work is often the thing that did not happen.

Losses that did not happen book nowhere. Ask for the record of them, rate the reasoning, and reread it when the outcomes come in.