Remediation is capacity, not an exceptional item

Every few months an Indian generics company announces that a USFDA inspection has gone badly. The quarterly results that follow have a familiar shape. Consultants' fees, retesting, additional staff and write-offs of batches on hold are grouped together and presented as remediation costs. Analysts add them back. Management speaks of a temporary disruption. A year or two later, if the site clears, the costs vanish from the commentary as though they had been weather.

I have presented results in that shape myself, and I now think the presentation does quiet harm. The costs are real and the disclosure is usually honest. The harm lies in what it teaches everyone involved, including the board: that quality is an interruption to the business rather than one of its main inputs. A plant that the USFDA inspects is not a factory with a compliance department attached. Its ability to sell into the United States rests on its laboratory, its records and its people as much as on its tablet presses. When we call the cost of restoring those things exceptional, we imply that the earlier level of spending was adequate. The warning letter usually suggests otherwise.

The adjustment that teaches the wrong lesson

Consider what happened at our Unit III. The inspection in November 2021 found problems in the quality-control laboratory: audit trails that were not reviewed, investigations closed without adequate root causes, analysts carrying more work than they could complete carefully. None of it was sudden. It accumulated over several years in which production volumes grew faster than laboratory capacity, and in which every budget review, including reviews I attended, treated quality headcount as overhead to be held flat.

When the warning letter arrived, we spent ₹285 crore over two years to put things right: ₹190 crore of operating cost and ₹95 crore on laboratory systems and equipment. We also ran the plant about 20 per cent below its previous output for eighteen months. A meaningful part of that spending simply restored what the plant should have been spending all along. Had we presented all of it as exceptional, we would have told investors, and ourselves, that the earlier margins were the true margins. They were not. They were margins earned by under-investing in the function that permits us to sell.

The market's habit of adding back remediation creates an incentive nobody designed. A finance team that holds quality spending down in ordinary years reports better recurring margins. If an inspection later goes badly, the catch-up is labelled exceptional and largely forgiven. Over a full cycle, the company that under-invests and then remediates can look better on adjusted measures than the company that spends steadily and is never cited. That is the wrong lesson, and boards absorb it without anyone intending them to.

I am not arguing that every rupee of remediation belongs in recurring cost. Consultants engaged for a fixed programme, one-time retesting of retained samples and the write-off of batches that cannot be released are genuinely non-recurring. But the additional analysts, the data systems and the training that the plant needs permanently are part of the cost of running it. Our 2022–23 accounts separated the two. The separation was more useful to directors than either a single exceptional figure or no distinction at all, and it gave analysts a base from which to judge whether our later margins were sustainable.

Measure the capacity you can release

The deeper problem is that our planning language describes production capacity and very little else. Budgets state how many tablets each site can compress, coat and pack. Few state how many batches the laboratory can test, review and release at normal staffing without cutting corners. Yet in a regulated plant the second number is the binding constraint more often than the first.

When we rebuilt Unit III, we began measuring release capacity explicitly: analyst hours available against the testing load implied by the production plan, how long open deviations and investigations have been waiting, and the share of batches released within the standard review time. These figures now sit in the monthly management report beside production volumes for every site. The first time we presented them for all four plants, they showed that Unit II, which had never been cited, was running its laboratory at a utilisation that would have alarmed us at Unit III. We added staff there before an inspector had to point it out.

Release capacity also changes capital decisions. A proposal to add a compression line now has to show the laboratory, warehouse and quality-assurance resources that the added output will need. Several proposals have become more expensive under that test, and one was deferred. That is the point. An expansion whose output cannot be released carefully is not an expansion. It is a future observation waiting to be written up.

What I now ask a plant budget to show

My requests of plant budgets are now specific. I ask for laboratory staffing against the testing plan, not against last year's headcount. I ask for the backlog of deviations and investigations, with their ages, treated like a balance that should fall rather than a list kept inside the quality department. I ask for the cost of data-integrity controls, including system validation and periodic audit-trail review, as a recurring line. And I ask each plant head what output the site could sustain if the laboratory were staffed exactly as planned and not one analyst more.

Directors can ask the same questions without becoming quality experts. They need not judge whether a particular investigation was adequate; that belongs to qualified people. They can ask whether release capacity is planned, whether the backlog is rising or falling, and whether adjusted earnings rely on treating permanent costs as temporary. An Audit Committee that asks those questions every quarter will learn more about inspection risk than one that waits for the next disclosure to the stock exchanges.

There is a cost to this approach, and I should be clear about it. Our recurring margins are lower than they would be if we presented quality spending the old way, and some investors have asked why our margins trail those of peers with similar portfolios. My answer is that I would rather defend a margin that includes the cost of being allowed to sell than one that borrows from the next inspection. So far the board has agreed with me. The licence to sell from a plant is never recorded as an asset, but it is the one we would miss first.