A credit agreement gets one serious reading. It happens in the weeks before close, by the deal team and their counsel, with a markup open on one screen and the term sheet on the other. Every basket is argued over. Every definition is traced. By signing, three or four people know the document better than anyone ever will again.
Then it is filed. From that day, the document is a reference work that nobody has time to consult, and the questions that need it start arriving.
The questions that send you back to the PDF
They come from the desk, from portfolio, from risk and from the borrower, and they have a family resemblance. Can they incur another 150 under the general debt basket, or has the ratio test to be met? What is left in the restricted payments builder after the dividend in March? Is this disposal permitted or does it need consent? What is the leverage headroom at the next test date on the numbers in the compliance certificate? Does the change of control definition catch a secondary sale to a sponsor affiliate?
None of these is hard in principle. Each has an answer in the agreement. But answering means finding the covenant, following its defined terms back through the definitions section, checking the amendments for anything that moved, and doing the arithmetic against figures that live in a separate compliance certificate. For a large-cap document that is an hour for an analyst who knows the deal and half a day for one who does not.
The result is predictable. The question gets answered from memory, or from the term sheet, or from the summary in the credit paper, all of which were accurate at close and none of which reflect the amendment in year two.
What is actually missing
It is tempting to say the missing thing is a summary. Every bank has summaries. Deal teams write them, credit papers contain them, and law firms produce very good ones at close. A summary is a snapshot, and it has the same problem the document has: it is read once.
The missing thing is a structured, current answer to the standing questions. Not "what does section 7.02 say" but "what is the capacity under 7.02(b)(viii) today, given the last certificate and the last amendment, and where in the document does that come from". That is a different object from a summary. It has to be computed, it has to hold state between test dates, and it has to cite its sources so that the person relying on it can check.
A document read once at close and never again is not a control. It is a record of what the control was meant to be.
Why spreadsheets do not close the gap
Most portfolio teams have a covenant tracker. It holds the maintenance tests, the thresholds and the last reported ratio, and it does a reasonable job on the question it was built for. It does not hold the incurrence package, because the incurrence package is not a list of numbers; it is a set of interlocking definitions and baskets that only produce a number when you ask a specific question at a specific date. Nobody keys that into a spreadsheet, so the spreadsheet cannot answer it, so the PDF gets opened again.
The tracker also decays. It was built by someone for one deal, copied to the next, and edited by whoever inherited it. The threshold in the cell is the one from close. The amendment that stepped it down is in a folder.
The standard we think is right
We are not going to name a product here, because the point holds whether you build or buy. Any credit team with more than a handful of documents should be able to answer the standing questions without opening the PDF, from a structure that was populated from the executed documents, updated when amendments and certificates arrive, and cited to the page on every number.
The document still gets its one serious reading at close. What changes is that the reading produces something that keeps working after the readers have moved on to the next deal.