पाठशाला Pathshala · विचार Vichār, The idea · Lesson 20 · Build

Regulatory arbitrage ideas and their expiry date

Some companies earn their margin from a product and some from a rule. The second kind can be a fine business for a while. Know which one you run and date the rule.

Pathshala, The Founder Library · 11 October 2026 · 8 min read

The stone instruments of the Jantar Mantar observatory in Jaipur in bright sunlight. — Photograph: Sanket Sawale · Pexels

Some companies make money because their product is better. Others make money because a rule lets them do something the incumbent may not, or does not yet forbid something it soon will. The second kind can be a perfectly good business for three or four years. The founders who get hurt are the ones who never noticed which kind they were running.

This lesson gives the difference a test, sorts rule edges into four shapes, walks through two Indian rules that changed on a date, and turns the whole question into a number you can put in the plan: the expected life of the edge and the margin left on the morning it goes. It closes with a register to keep every quarter.

Product edge or rule edge

Ask one question of the gross margin line. If the rule changed tomorrow so that you were treated exactly like the incumbent, taxed the same, licensed the same, classified the same, what share of this margin would survive? The part that survives is your product edge. The part that vanishes is your rule edge. Most companies have some of each. A payments company built on zero merchant fees has a rule edge it shares with everyone; a lender that underwrites better than the bank has a product edge no rule can take away. The question is the proportion, and the honest answer is usually higher than the founders would like.

A rule edge is not a moral failing. Rules are part of the market and reading them is part of the job. Uber’s own prospectus in April 2019 listed among its summary risk factors that its business would be adversely affected if drivers were classified as employees instead of independent contractors. That is a company stating, in a filing, that a share of its margin rests on a classification. The mistake is not having the edge. The mistake is building the cost base, the valuation and the fundraising story as though the edge were permanent.

Four shapes a rule edge takes

The gap. The regulator has not yet written a rule for what you do, so you do it without the cost of compliance that a regulated competitor carries. Gaps close fastest of all, because the regulator is usually already drafting. The tell is a consultation paper or a working group with your category in its title.

The exemption. A rule explicitly carves you out: a tax holiday for a defined period, a threshold below which a filing is not required, a scheme for a class of company. Exemptions are the kindest rule edge because they usually carry a printed expiry, and the only mistake available is not reading it.

The classification. The same activity is treated differently depending on what it is called: skill or chance, marketplace or retailer, contractor or employee, software or financial product. Classification edges last as long as the line holds and end abruptly when a court or a legislature redraws it, often for the whole category at once.

The jurisdiction. The business is done from or routed through a place where the rule is different. These edges end when the stricter jurisdiction reaches out, through payment rails, app stores or tax treaties, and they tend to end for everyone on the same day.

How rules expire: two Indian cases

Real-money online gaming in India spent years on a classification edge. Games of skill were treated differently from games of chance, and an industry built itself on the line between them. On 20 August 2025 the Promotion and Regulation of Online Gaming Bill was introduced and passed in the Lok Sabha, and the Rajya Sabha passed it the next day. As PRS Legislative Research summarises it, the bill prohibits offering online money games, defined as games where users pay money or stakes in the expectation of winnings, whether the outcome depends on skill, chance or both. Offering them carries up to three years in prison and a fine of up to ₹1 crore; advertising them up to two years and ₹50 lakh; facilitating payments for them up to three years and ₹1 crore. The single word that had held the industry up, skill, was written out of the test. Two days in Parliament took away a classification that had held for years.

A colonnaded government building in New Delhi with the Indian flag against a bright sky. — Rules are written in buildings like this one and they change on a date. The companies that read the drafts were ready for it. Photograph: Abhinav Tripathi · Pexels

Digital lending is the slower and more typical case. Before 2022 many lending apps earned fees charged to the borrower by the app itself, routed money through their own accounts and offered partner lenders a first-loss guarantee that let the app take credit risk without holding a lending licence. RBI’s guidelines of 2 September 2022 required loans to be disbursed into the borrower’s own bank account and never through the account of the app or any third party, made fees to lending service providers payable by the lender rather than the borrower, demanded an upfront annual percentage rate and a key fact statement, and put first-loss guarantees under the 2021 securitisation directions. Those guidelines have since been repealed and replaced by the Digital Lending Directions of 8 May 2025, which is itself a reminder that the rule keeps moving after the first change. Here the rule did not ban the business. It removed the parts of the margin that came from the gap and left the parts that came from the product.

The clock: pricing the rule into the plan

A rule edge is easier to manage once it has a number. Take the share of gross margin the rule creates and the chance, in any given year, that the rule changes. A one-in-four chance a year gives the edge an expected life of four years and roughly a 42 per cent chance of still standing in three. Multiply through and the five-year plan has a smaller number at the bottom than the spreadsheet that assumed the rule would hold.

Take a lending app in 2021 with ₹8 crore of gross margin, of which ₹4.4 crore is platform fees charged to borrowers and the rest is the spread on loans it originates for a partner NBFC with a first-loss guarantee. The fee line is a gap edge: 55 per cent of margin. Put a one-in-four annual chance on the rule changing and the plan should expect about 70 per cent of the five-year margin it shows, with ₹3.6 crore a year left on the morning the rule arrives. That company should have been spending its fee income on the underwriting model and the borrower relationship, not on a cost base sized for ₹8 crore forever. The figure lets you put in your own numbers.

A rule edge is a loan from the regulator. Write down the repayment date.

Turning a rule edge into a product edge before the date

The right use of a rule edge is to buy time to build something the rule cannot take away. Four conversions work. Customers: the edge funds acquisition; the relationship outlives the edge if the product is good enough that customers stay when the price moves. Data: years of transactions inside the window become an underwriting model, a demand forecast or a fraud signal that a late entrant cannot buy. Licence: the cleanest move is often to become the regulated thing, an NBFC, a payment aggregator, a licensed operator, before the rule forces it, so that the day the gap closes is a day your competitors lose an edge and you do not. Cost: an operation built to make money at the post-rule margin, run during the window at the pre-rule margin, throws off cash.

The conversions that fail are the ones that spend the edge on things that only make sense while it lasts: a cost base sized to the rule-inflated margin, a valuation priced on it, a fundraising story that treats it as a moat. An investor who has seen a rule change before will ask what the business earns without it. Answer that question in the deck before it is asked, and the rule edge reads as judgement rather than luck. The [moats lesson](/library/moats-that-matter-in-year-one) covers what a durable edge looks like once the rule is gone.

Signals that the clock is running

Rules rarely change without warning. The signals arrive in a sequence, and each one should raise the annual probability in the figure. A regulator’s speech or annual report names the category. A working group or committee is formed. A draft direction or consultation paper is published for comment. Complaint volumes rise, or a court case reaches a High Court and then the Supreme Court. The incumbents start to lobby in public. The press coverage turns from the founders to the customers who were hurt. By the time the final rule is notified, the companies that read the sequence have restructured; the ones that did not are writing to their investors.

Watch the global version too. A rule that has changed for a similar business in another large market is a strong signal, because regulators read each other. And watch your own margin: if the share coming from the rule is rising year on year, the business is becoming more dependent on the edge, not less, which is the opposite of what the window is for.

A quarterly rule register

Keep one sheet. One row for each rule your margin touches: what the rule is, which of the four shapes it takes, the share of gross margin it creates, the regulator, the last signal and its date, the annual probability you assign, and what you are building that would survive the change. Review it on the first working day of each quarter alongside the board pack. Raise the probability for every new signal; lower it only when a rule is confirmed in law with a long horizon. If any single rule creates more than half of gross margin, the board should see the row, the conversion plan and the date by which the plan makes the rule irrelevant. The [why-now lesson](/library/why-now-timing-argument) is the other half of this sheet: the shift that opened your window, beside the rule that will close it.

Nothing here is legal, tax or investment advice. Rules quoted were checked in October 2026 against the sources below; read the current text with counsel before acting on any of them.