A startling fact
If you had invested in the top 20 Indian business groups of the 1960s, more than half would have disappeared by now. Names that once dominated—Thapars, Scindias, Mafatlals—have faded into history. Corporate mortality is real, yet most investors obsess over how firms succeed rather than why they fail. But as someone in the credit business, I care about one thing, that is avoiding the left tail of the distribution—bankruptcies.
Bankruptcies are a routine global phenomenon. Global default rates hover around 2% pa. They rise to 3.7% for speculative-grade loans/bonds. In India, the 10-year average corporate default rate is 2.5%, but in the mid-2010s, it spiked to over 10%. Today, it’s below 1% pa. But bear in mind that these are annual bankruptcies. Once you take a long-term view – two third of small private firms go extinct within a decade in US. Even for old firms – say ones which are listed in US, about 15-20% go bankrupt every decade. In India too – 15-25% of the firms go bankrupt every decade. One quick lesson for everyone here: Credit (same applies to equities) as an instrument cannot be for the long term buy and hold. You must ensure that you have an oversight on the evolving condition of the firm – lest it decays and go bankrupt. The adjacent point is that credit is also – a monitoring heavy business. As investor, you must ensure that firm takes right measures to ensure that its decay isn’t in motion.
Where do these failures cluster? In India, 60% of defaults have come from just five sectors—steel, power, telecom, infrastructure, and textiles (2017 banking data). Telecom, however, consistently stands out across the globe. 25% of global defaults were in telecom in 2024. The latest industry joke? Telecom is the new airline business—perpetually struggling, always on the brink, and a graveyard for capital every time the cycle turns. There are many other sectors as well – which are extremely cyclical and are prone to frequent bankruptcies. Retail, Restaurants, and new tech businesses are some such ones.
So, what makes some firms destined to fail while others survive? And more importantly, how do I assess which firms are worth lending to? That’s the objective of researching at Centre for Finance research at IIM Udaipur. Some of the insights that I share in this paper comes from case studies of a few dozen firms that we researched at the institute.
First a bit about frameworks of assessing credit worthiness. Some believe that a firm’s creditworthiness can be determined solely by leverage and coverage ratios, making risk assessment appear straightforward. It’s a clean, mechanical way to evaluate risk. Bansal, Chopra, and Wadhwa’s research backs this up. Their logistic model found that bankruptcy risk could largely be predicted by a few financial indicators.
“For two years prior to a bankruptcy event, the log of the odds of a company going bankrupt was negatively related to return on assets (p = 0.0012) and current ratio (p = 0.0243), while positively related to the ratio of working capital to total assets (p = 0.0176) and the ratio of sales to total assets (p = 0.0103).” https://arxiv.org/pdf/2008.04782
In Plain English
Higher return on assets and current ratios reduce bankruptcy risk, while high working capital and sales relative to assets indicate a higher risk. One year before bankruptcy, the same relationships hold, but with even stronger statistical significance. This will seem obvious to practitioners. But ask a bunch, you will rarely get this clear an answer.
Anyways, this model has an accuracy claim of over 80%. And yet, something about it feels incomplete. If risk were purely about a set of financial ratios, lending would be as easy as plugging numbers into a spreadsheet. But any credit analyst who has been through a few cycles knows that firms don’t fail just because a ratio crosses a threshold. Context matters.
I am a relativist and probablist – My model entails looking at the firms relatively and assign probabilities to various important vectors – if they could drive firms bankrupt.
Let me explain. The key relativist’ Q in my mind is- how is the firm doing relative to its own history and viz competition. I think that the stress emerges when a company’s financials deteriorate over time and, more critically, when it starts lagging behind its peers. A declining firm in a strong sector is more concerning than one struggling in a weak industry.
Look at microfinance today—some firms are visibly under strain, not just in absolute terms but relative to their sector. My instinct? Avoid these alpha-negative firms. It’s usually safer to back firms suffering from a sector-wide downturn rather than those battling firm-specific issues. Sectoral weakness tends to reverse unless driven by structural shifts, such as product obsolescence rather than cyclical slowdowns. That’s why I’m particularly wary of SaaS businesses today—many are at risk of being replaced by AI-driven platforms.
The second part of being probablist is, while I evaluate the decisions to take a credit call on a firm – I primarily ask two questions from my colleagues and researchers, first, assessing its risk relatively— both in the context of the firm’s own history and its competition—and second, evaluating the probability of failure across a few critical dimensions.
Some of those critical dimensions are leverage, competition, commodity market downturn and import dumping, business cycle slowdown, and policies. How can one predict firm collapse in advance? The key is breaking down corporate risk into vectors or variables—specific indicators that drive distress. Here’s my list of core risk vectors—each a potential domino that can bring a firm down.
Vectors Probability
Competition: Price wars and new technology disruptions
Leverage: High debt levels
Promoter Leverage
Commodity Market Downturn
Business Cycle Downturn: Revenue slowdown, price pressures
Global slowdown (commodity cycle)
Regulatory Issues: Compliance and legal challenges
Government Receivables: Delayed payments from government entities
Governance: Corporate governance issues and fraud
Financial Cycle: Rollover of finance and asset-liability management issues
Poor Underwriting as Financiers: Ineffective lending practices
Poor Business & Strategy: Strategic missteps and operational failures, management competence
There are two key points to make here. First, in most failures, multiple factors are at play—it’s rarely just one. However, there is almost always a HERO factor, the dominant driver of distress. I view this probabilistically- to assess ex ante, which firms will go down. For example, when analysing an auto component manufacturer last week, I was less concerned about the cyclicality of the business, but the biggest risk in my mind was the potential dumping of Chinese auto components and the wave of EVs, that most Indian firms may not be ready for.
Work with IIM Udaipur’s centre of financial research
I gave my framework to students analysing past firm defaults. Professor Bhavya of IIM Udaipur, probabilities to all factors, even giving 1-3% likelihood to risks that weren’t even relevant. When I pointed at interesting mistaken than all the students and researchers made. They assigned asked why, the response was: “We thought we needed to apply all of these to every firm.” This is a classic analytical pitfall for most of us. On one side some of us obsess over just one factor – driving our view on the firm, and then – there are analysts who will not weight the risk appropriately, and end up being almost equal weight on multiple factors while deciding on a credit decision.
I once worked with someone who would produce a 200+ page report on a single firm—impressive in its depth but ultimately not very useful in practice. The sheer spectacle of such a report was one thing, but the real question was: What do I do with it? Net, net, not all risk factors apply to every firm. This framework isn’t a checklist—it’s a way to think about risk.
Okay – back to our work with CFR. Prof Bhavya asked students how did they assess probabilities across each vector. Researchers said – what it was – a purely subjective exercise. In the defence of this model, I argued that this was inspired by Tetlock’s super forecasting framework, one that has had a huge impact on how I think of various uncertain events. The way it works is -Split a big question into sum of parts, and then assign probability to each part. Even though it’s subjective, the aggregate view becomes a lot more objective than being driven by mono-variable.
Deepanshu, a senior student & researcher at IIM Udaipur, suggested an important improvisation in my framework. While evaluating a firm, instead of assigning a straightforward probability to each risk factor, he broke it down into three parts: Impact, Evidence, and Cascading Effect. He applied this approach while analyzing the case of DHFL, and I see this as an evolution of my framework. Rather than simply assigning a subjective probability, this method structures the assessment in a more rigorous way.
His core idea is simple:
- Assess the potential impact of each risk factor.
- Look for evidence—any data points or signals that indicate whether this risk might materialize.
- Evaluate the cascading effect—how this risk, once triggered, could influence other variables and amplify distress.
This structured probability assessment provides a clearer, more actionable way to think about why DHFL failed. Here is the way he assigned different ‘scores’ to each of these variables.
Proability of Reasons
Variable |
Impact |
Evidence |
Cascading Effect |
Probability |
|---|---|---|---|---|
|
Competition: Price wars and new technology disruptions |
0 |
0 |
0 |
0.00% |
|
Leverage: High debt levels |
8 |
10 |
8 |
15.12% |
|
Promoter Leverage: Financial exposure of company promoters |
2 |
8 |
2 |
6.98% |
|
Capital Market Leverage: Impact of share price fluctuations |
5 |
7 |
2 |
8.14% |
|
Commodity Market Downturn: Effects of margin cycles |
7 |
8 |
7 |
12.79% |
|
Business Cycle Downturn: Revenue slowdown |
0 |
0 |
0 |
0.00% |
|
Global Slowdown |
0 |
0 |
0 |
0.00% |
|
Regulatory Issues: Compliance and legal challenges |
3 |
10 |
2 |
8.72% |
|
Government Receivables: Delayed payments from government entities |
0 |
0 |
0 |
0.00% |
|
Governance: Corporate governance issues and fraud |
9 |
8 |
9 |
15.12% |
|
Financial Cycle: Rollover of finance and asset-liability management issues |
9 |
10 |
8 |
15.70% |
|
Poor Underwriting as Financiers: Ineffective lending practices |
10 |
10 |
10 |
17.44% |
|
Poor Business and Strategy: Strategic missteps and operational failures, management competence |
0 |
0 |
0 |
0.00% |
|
Total |
53 |
71 |
48 |
100.00% |
The Impact score (0-10) measures the severity of the issue on DHFL’s financial health and operations.
The total probability is 100%, meaning it distributes the contribution of each factor to the overall failure. The probability for each reason is computed as:
Where, in this particular case,
- The total sum of Impact, Evidence, and Cascading Effect is 53 + 71 + 48 = 172 (sum of all factors).
- Each factor’s contribution is divided by 172 and converted into a percentage. To get the probability which is assigned to each vector.
From Framework to Case Studies: Understanding the 2015-18 Corporate Mess Having lived through the horrid times of 2018’s financial crisis in India, I have often tried to ask what led to the 2015-18 corporate NPA mess? A bit of history is important here. By 2009, Indian firms were the most leveraged compared to any other country in Asia. This excessive debt led to a wave of corporate debt restructuring (CDR) between 2012-14. So when a cyclical downturn hit, commodity prices collapsed, and policy missteps compounded the situation, a wave of defaults followed across Indian industrials. But here’s the key point: Each insolvency had its own unique story. And it’s unfortunate that so little has been written about them.
Jairus Banaji makes a very interesting point about corporate bankruptcies of past few decades, “What drove firms to insolvency was some combination of strategic miscalculation (Mallya launching Kingfisher with little grasp of the airline industry; the Mafatlals buying out Shell’s stake in Nocil at considerable cost), formidable competition from global players (Videocon and BPL in consumer electronics when LG and Samsung were making their entry), and downright fraud (the siphoning of loans)—all coupled, of course, with gargantuan levels of debt that state-owned banks had no compunction about creating” https://www.phenomenalworld.org/analysis/family-business/
So with this context, I asked researchers at IIM Udaipur to find why various firms defaulted during last decade. Over next few months, I will write about each of the case in detail. Of JSPL, DHFL, ILFS, CCD, ALTICO, ADAG and 10s of other such cases. But here, I am writing on three things which were common across many of the bankrupt firms. Those were mismanagement, competition, and fraud.
Mismanagement: Over-Leverage, Unrelated Expansion, Promoter Leverage, family splits, Govt linkage
Firm failures stem from multiple factors, but two primary causes stood out across a 2 dozen high profile business failures that we studied. The first is the business or commodity cycle, the most common reason of large scale bankruptcies in developed markets. The second—and far more prevalent among Indian businesses—is mismanagement by promoters.
There are many lessons to be learned. Aayush, one of the researchers wrote on the case of Café Coffee Day “Founder and CEO V.G. Siddhartha aggressively expanded and diversified the business using debt. While his ventures in hotels, logistics, tech parks, and financial services performed relatively well, he funnelled their cash flows into the café business to support rapid expansion. Unfortunately, this expansion failed to generate sufficient revenue to service the mounting debt. As a result, financial distress spread across the entire group, leading to its collapse.
Similarly, Vijay Mallya’s United Breweries (UB) Group, which had interests in breweries, took a risky turn when he entered the aviation sector. He launched Kingfisher Airlines as a luxury carrier but later acquired Air Deccan, a budget airline—completely contradicting Kingfisher’s business strategy. This misalignment, coupled with excessive debt, turned Kingfisher into a non-performing asset. When the airline collapsed in 2013, UB Group was forced to sell off its profitable businesses, including chemicals, fertilizers, and spirits, to repay debts.
…..overleveraging and resource misallocation—often driven by overambitious promoters—can lead even once-thriving enterprises into financial crisis.”
The story such that of CCD, and Kingfisher appear again and again across large business failures in India. Passionate and successful promoter – levers up significantly, and enter in unrelated businesses or expands too rapidly. The most interesting part is that almost until the very end, in most such cases we found that promoters thought they would pull off, and that’s why they gave personal guarantees, PDCs and many other personal collaterals to save their failing franchises.
One more factor, that recurs across Indian firm history, and is worth keeping in mind: family splits. Whenever a business family divides, at least one faction often ends up in a significantly weaker position. Asset allocation decisions by one part, is often made hastily or without prudence, ends up being disastrous. Does ADAG, Thapars, Goenkas, or Jindal’s come to mind?
One more point on promoters. For many in my generation, one hard-earned lesson is to be sceptic of sectors that require government licenses to operate. If your business model depends on government payments, guarantees, or receivables, think real hard.
I learned this first-hand when IL&FS collapsed. Both Jharkhand road projects—which had state government guarantees—and IL&FS Education, with its receivables from various state governments, defaulted to my portfolios. The good news? Eventually, everything got sorted. The lesson? Never expect timely government payments, and don’t rely on guarantees to be honoured in a timely manner. The right way to assess the Govt related transactions is this, that delays are more common than what investors are ready for, but the recovery rate is also high. So it’s a bad AAA/AA investment. Perhaps a good A/BBB one.
Competition
A few firms that were researched by students– died off stiff competition. The most interesting sector that we studied was telecom, that saw competition driving most firms extinct. First, too many players were allowed to enter, then spectrum was priced excessively and eventually, JIO’s entry killed almost all the players in the businesses.
I was gullible enough to finance Vodafone in 2017. The good news? We recovered all the money. The bad news? Looking back, I wonder what we were smoking to not see the scale of Jio’s disruption. I am wiser today – and that’s why I cautioned Asian Paint investors 2-3 years ago – about what they were getting in, when Grasim was entering in the pain business. The good news? I wrote in a blog post- “AP won’t go bankrupt—it’s well-run and lightly leveraged. The bad news? It’s no longer a price maker” Ditto is my caution for AMCs now. The Jio financials and many such firms will likely start the true ETF evolution in India. Thus both the margins and the pace of growth will begin to dwindle in bottom-up equity franchises.
Governance issues and misrepresentations
Our researchers studied both JSPL and Bhushan Steel—which defaulted to lenders. Yet, financiers barely lost money in JSPL, while Bhushan’s default resulted in massive losses. I asked Kumar Shakhapure why. His researched answer was: luck (a steel price upswing) and side pockets (other assets to sell, family support). I disagreed though. I believed the key difference was governance—JSPL didn’t have any cash leakage. Once the cycle turned, it could solve its problems. Bhushan Steel, on the other hand, suffered from serious governance failures and alleged fraud. Promoters allegedly diverted loans into personal businesses and properties, rather than reinvesting in the company.
Cash leakage is a key form of governance lapse and is a chronic problem across firms which failed over past two decades. One of the many reasons why India recovery rates are less than 30% even when in OECD it’s close to 70%. Most analysts (including me) rely on traditional methods to detect governance issues
- Meeting management, and reading between the lines. The most interesting part about humans is that it’s very difficult to lie in person. “Zoom” conversations will never reveal all that to you. So don’t be lazy. Go and sit with the management for a few hours, and you will figure.
- Speaking with vendors, ex-employees, and competitors: More we talk to different people in the ecosystem that the firm operates in, clearer insights we get.
But is there a quantitative way to detect fraud? I asked Shivam Tayal for insights. He wrote in his case study, “Corporate fraud can be understood through the Fraud Triangle, which identifies three key drivers:
- Pressure (financial stress forcing unethical actions)
- Opportunity (weak internal controls allowing fraud to occur)
- Rationalization (justifications for unethical behaviour)
To detect fraud using numbers, forensic experts rely on two key models: The Beneish M-Score and the Altman Z-Score.
- The M-Score flags earnings manipulation by analysing financial ratios like revenue growth, asset quality, and expenses. A score above -2.22 suggests possible fraud.
- The Z-Score predicts financial distress based on profitability, leverage, liquidity, solvency, and turnover. A score below 1.81 signals high bankruptcy risk, while a declining Z-Score over time can indicate financial manipulation.
While these models have limitations, using them together provides a more holistic fraud detection framework. The M-Score helps identify accounting manipulation, while the Z-Score acts as an early warning system for financial distress. When combined with traditional forensic assessments, these tools can help us detect risks before they escalate.
For listed firms, I think the equity price becomes a reasonable metric of detecting ‘strange things’ in the firm. Company’s equity is after all a call option on firm’s assets. And is also seen by many – most importantly insiders, to profit, if something dramatically wrong is ongoing there. As such, I find equity a lot more “ear to the ground asset” than credit.
I have to admit – I don’t use quantitative methods extensively. But I will now. I am also quite bullish about the role of AI agents in doing a lot of it. In fact – we already have an agent that works ‘sincerely’ to give us a good sense of if there is a risk of misgovernance in the firms. These agents will ‘love’ if I actually given them a good quantitative framework to work on. Visit this page again. I will share lots of insights from case studies. But also – how my AI agents are performing.
