Not All Governance Scares are Created Equal

On 19 March 2026, HDFC Bank lost more than 1 Lakh Cr of market value in a single session. The then chairman, Atanu Chakraborty, resigned citing "incongruence between his personal values and the bank's practices." He gave no further details.

Nothing else had changed. The deposits were where they were the day before. The loan book hadn't moved. The net interest margin (NIM) was the same. A man had quit a job, with a cryptic sentence, and the market took a number as large off the table.

Two months later, on 27 May 2026, the Indian Express reported that the bank had paid 45 Cr in disguised interest to a Maharashtra state agency, routed through marketing expenses to dodge RBI's master directions on deposit rates. Same bank. Possibly a more concerning allegation. However, the stock fell just over 2%, which was approximately 24,000 Cr of market value, a big number, sure, but roughly a fifth of what had happened a few months back.

Same company. Same year. Two governance headlines. Two very different reactions.

If you're a long-term investor, and bad news about corporate governance hits your portfolio, the more important question becomes what do you do. Sell immediately? Wait and see? Ignore it if the earnings are good?

The honest answer is that it depends. And, for the same reason, we pulled 20 years of NSE data, picked 22 of the governance events since Satyam in 2009, and ran the numbers. What we found is that the kind of bad news matters far more than how bad it sounds. And that the market's first reaction is sometimes wildly wrong, in both directions.

The Suspicious Verdict

It is essential to read governance news for what it is—an update rather than a verdict.

On 24 January 2023, a US short-seller named Hindenburg Research released a 100-page PDF accusing the group of stock manipulation and improper use of tax havens. Within seven trading sessions, Adani Enterprises was down ~70% with a ripple effect across the group.

By 27 May 2026, three years and four months later, the Adani listed companies had not only recovered but were above the pre-Hindenburg level.

The business in between? Largely unchanged. Ports still moved cargo, power plants still generated electricity, the airports still ran. What changed was the market's probability estimate of the worst-case allegations being true.

By 2026, after a SEBI probe that found nothing material, a Supreme Court that declined further investigation, and a US DOJ that quietly walked away from criminal charges, the worst-case probability had then collapsed.

The market wasn't pricing the news in 2023. It was pricing the worst-case implications of the news. When those implications didn't materialize, the price came back.

This is what we mean when we say governance news is a probabilistic update, not a verdict. Once you see it this way, the field gets a lot clearer.

All Governance Events are Not the Same

Lumping Satyam with the HDFC chairman resignation did not make any sense. So, we tried to split these events into roughly seven kinds of governance shocks.

Confirmed fraud: The books are fake or the cash just isn't there.
Few examples would be Satyam in 2009, Manpasand in 2018, Vakrangee, also 2018. This news is binary and verifiable, once it's out, it's out, and the recovery, if any, is rare.
Promoter integrity: Related-party transactions, criminal conduct.
Yes Bank under Rana Kapoor, ICICI's Chanda Kochhar, the Singh brothers across Religare and Fortis are some of the examples.
Regulatory or legal probe: An investigation by SEBI, SFIO, the DOJ, the SEC.
Adani's US bribery allegations (2024-26) is an optimal example. Usually multi-year resolutions with wide outcome ranges.
Short-seller, activist, or whistleblower report: A third party makes detailed allegations; it is not necessary that the company gets charged. Mixed credibility could often lead to bimodal outcomes.
Board or structural governance: An independent director resigns, the chairman exits, the auditor walks away.
Atanu Chakraborty's HDFC Bank resignation in March 2026 is an example of this.
Related party or disclosure lapse: A specific transaction wasn't disclosed correctly. Not always fraud, sometimes a technical lapse, sometimes the first thread of a larger problem.
PC Jeweller in 2018 or HDFC Bank's MSRDC deposit story in May 2026. Usually narrow.

Most would instinctively put HDFC's chairman exit and Adani's DOJ case in the same bucket. They're not. One is Type E. The other is Type C. The data, as we'll see, treats them very differently.

The Data: 22 Events, 20 Years

We took the data going back to 2006 and ran each of the 22 events through the same template:

Find the closing price the day before the event
Track the stock day by day for the next five years (or until the data ends)
Compute the drawdown relative to Nifty 50 (because a 30% fall in March 2020 isn't a governance event, but a much broader systematic fall)
Note the trough, count the days to recovery, and bucket by PE, ROE, and market cap at the time of the event

Here's what the data says, summarized by type:

Recovery from governance event by type

The headline finding is in the recovery column. Type A and Type B events kill more than half the companies they touch, while every single one in our Type C bucket, eventually recovered and Type D and Type E are a 50:50 story (small sample size).

Read that again. Every regulatory probe in our sample eventually saw the stock recover.

That includes Adani (twice—Hindenburg and the DOJ case), Sun Pharma's 2018 whistleblower episode, Edelweiss after the RBI ban, and Ranbaxy (whose "recovery" technically came through Sun Pharma's acquisition). These were all multi-year stories; none went to zero.

Compare that to Type A. Of the five accounting frauds in our sample, three never recovered—Satyam, Manpasand, Vakrangee. CG Power was rescued by Tube Investments at a steep discount, so it counts as a recovery, but original shareholders took a massive drawdown along the way. Brightcom's recovery is fragile, it's technically positive but the stock remains a shell of its former self.

Type B is even more grim if you account for the longer time horizon. Median recovery for the ones that did recover was 1,018 days which is almost three years. Yes Bank, DHFL, Coffee Day, Religare. These weren't shocks, they were slow unraveling.

The other insight is in the median one-year return. Only one bucket, Type D, the short-seller and whistleblower events, sits in positive territory. Two events don’t make up a large sample, but both the names in that bucket saw their underlying businesses getting cleared and their stocks fully back to the index. When third-party allegations don't survive scrutiny, the stock often recovers all the way.

Type E (board/structural) sits modestly below zero at -9%, and that median is partly an artefact because the HDFC Bank chairman exit of March 2026 is too recent for one-year data to have played out, so it counts as "not recovered" in our sample by default. Tata Motors and Tata Steel had recovered most of their losses within a year, which is more in line with what you'd expect for Type E.

(The cleanest example: ICICI Bank during the Chanda Kochhar episode. The stock fell a touch under 7% relative to Nifty, recovered within 130 days, and was up 36% relative to the index by T+730. The bank was fine. A senior executive had a conflict-of-interest problem. Markets eventually separated the two.)

The Expensive-Stock Trap

We sliced 18 of the 22 events by its PEs at the time of the event. Three buckets: cheap, mid-priced, and expensive.

The Expensive-Stock Trap

Expensive stocks fell more than twice compared to cheap or mid-priced ones. And a year later, they were down 49%, while the other two buckets had drifted back (-11% and -3%).

We think it's mechanical. For example, a stock trading at 80 PE is priced for many years of perfect execution. The moment governance news creates doubt about that execution, the multiple compresses. Vakrangee was trading at a PE of 89 when the audit issues surfaced. The stock didn't just absorb the bad news; it also lost the optimism premium baked into that multiple.

Also, "Cheap" doesn't protect you either, the cheap bucket was down ~11% at T+365. The cheap bucket is dominated by companies that were already in some trouble, which is why they were cheap. Yes Bank trading at a PE of 17 in September 2018 wasn't expensive, but the cheapness reflected market suspicion that wasn't yet fully priced.

The honest read of this table isn't that any one bucket won. All three ended below zero at T+365. It's more about the asymmetry. The downside on bad news is worse than the upside on no news.

Is "High Quality" a Disguise?

Here's the most uncomfortable finding in the data.

We expected high-ROE stocks, (the "quality" companies) to absorb governance shocks better. The thesis was that real underlying profitability would cushion the blow.

However, the data said something very different. Sorted by ROE at the time of the event, the high-ROE bucket performed worse, not better.

The "High Quality" Disguise

This made no sense to us until we looked at which companies were in the high-ROE bucket. Satyam, in January 2009, was reporting ROE of 26%. Vakrangee, before the audit broke, was reporting ROE of 30%.

These weren't real ROEs. They were the output of the fraud itself.

When a company is cooking its books, the reported quality metrics like Return on Equity (ROE), Return on Invested Capital (ROIC), and profitability in general is a larger part of what's being cooked. High reported ROE at the time of a governance event is not safety, it could be a disguise, too. Not necessarily that's always the case, but it could be.

To know more about such financial shenanigans, here is a timeless series by Capitalmind Premium, which we published back in 2019.

The Size Question: Does Market Cap Matter?

The original framing of this piece had us hypothesizing that large-caps would absorb governance shocks better than small-caps. More FII liquidity, more coverage, greater ability to raise emergency capital.

Does Market Cap Matter

Large caps did fall less on the median. But only half recovered. Small-caps fell harder, but more than two-thirds recovered. (The mid-cap sample is too small to draw conclusions, it's just Vakrangee.)

Please note that it excludes the small caps that went to zero before our data picked them up. Survivor bias is alive and well in this sample.

The cleaner read would be that market cap size is a weak predictor of governance shock recovery. The type of event predicts much more than the size of the company.

What Does This Mean for You?

Let's say Bank X, in your portfolio, drops 14% on a Thursday morning because the chairman has resigned with a vague public letter. What do you do?

The temptation is to either panic-sell or aggressively buy the dip. Both are wrong. Both miss the structure of the decision.

The framework that falls out of our data:

First, classify.

What type is this? In the HDFC March 2026 case, it's a Type E with possibly some Type B undertones (the "incongruence with personal values" line is the worrying part).

Second, ask five questions.

Is the allegation about the books or about behavior? Books are dangerous, Behavior, much less so.
Is there documentary evidence in the report?
Has any independent body confirmed any part of the allegation?
What is the promoter pledge level? High pledge plus margin calls could turn a 20% event into an 80% event.
Does the business still generate cash if the worst part of the allegation is true? If yes, the stock might eventually recover. If not, it might not.

Next, watch for the follow-ons like auditor resignation, sudden top management exits, credit rating actions, etc.

For most investors, the right action is to do nothing for the first 48-72 hours, gather information, classify the type, check the follow-on signals, and act with the structure. Information asymmetry is at its peak in the first 48 hours after a governance shock. Almost everyone who trades in that window, except insiders and forensic analysts trade on incomplete information.

How We Think About This at Capitalmind

No governance framework is foolproof. We'll come to that. But here's what we actually do.

The exception list

Before any stock is considered for a Capitalmind stock Investment Approach (Surge India, Adaptive Momentum, Apex India), it has to clear what is more of an exception list, a blacklist of names we won't hold regardless of how attractive the numbers look.

A stock typically lands on it for one of the following:

Group structure complexity, related-party transactions that aren't clearly arm's-length, opaque subsidiaries.
Any recent SEBI or regulatory action of substance.
Frequent auditor changes without a clean explanation.
High Promoter pledge or frequent dilution.
Cash conversion ratio persistently below our comfort band, reported earnings not supported by actual cash (depends significantly on the nature and phase of the business).

For stocks already in our portfolios, we run the same signals on a continuous basis.

What this framework cannot catch

No screening, ours or anyone else's, would have caught Satyam in advance. When the books are fictional, when cash balances are imaginary, when auditors are complicit, there is no external signal that reliably surfaces before the confession.

This is where position sizing takes prudence.

If a stock is 4% of your portfolio and goes to zero, you lose 4%. Painful, yes but recoverable. If the same stock is 25% because you fell in love with the story, you lose 25%.

Capitalmind IAs caps individual positions well below the level at which any one stock can inflict permanent damage. The screening reduces the probability of a Satyam slipping through. Concentration limits reduce the consequence if it does.