Organisational Forms, Corporate Governance, ESG and Capital-Budgeting Completions

Finance-Plus ring · Branch EC · Node EC1.07 · ~16 focused hours, mastery gate >= 85%

Nine million shares out of fifty-five carry 66.18% of the votes at Northbridge Media Corp. (synthetic). The people holding the other forty-six million own 83.64% of the cash flows and decide nothing. If the founder group ever wants liquidity it can sell 4.0 million of its shares, 44.44% of what it holds, fall to a 9.09% economic stake, and still command exactly half the vote.

That arrangement is lawful in the United States and lawful in India. It was created by a choice of organisational form, and an organisational form is nothing more than a rulebook about who is allowed to decide what, who bears the loss when the deciding goes badly, and who the tax authority bills.

The rulebook matters to an analyst for one reason. Limited liability plus a tradable share is what makes it possible to own a business you do not run, and the moment ownership separates from control, the person running the business has an interest that is close to yours but never identical to it. Every governance mechanism ever invented, from an independent audit committee to a debt covenant to a hostile bid, is an attempt to narrow that gap. None of them closes it.

A governance failure almost never arrives labelled as one. It arrives as an unsecured loan to a group entity, repayable on demand, with no schedule stipulated. It arrives as an auditor who resigns in the week before results. It arrives as a board of eleven with four independent directors when the listing rules required six. What the analyst has to do is turn that evidence into a number, defend the number's size, and keep it separable enough that a reader who disagrees can switch it off and see the valuation without it.

The environmental and social material has exactly the same shape and is usually taught without it. A carbon compliance cost is a line in the cost stack; an abatement project is a capital-budgeting decision with a payback and an internal rate of return; a water-scarce plant is a capacity assumption. Where an environmental or social factor does not reach a line in a model, it is not analysis and it should be labelled as something else.

And the projects that a board approves get judged by decision rules older than any of the disclosure regimes, several of which are wrong in ways that are easy to state and easy to demonstrate. A payback rule accepts a project worth minus $13,223 and rejects one worth plus $36,274, using the same three-year cutoff, on the same day. Two machines with different lives can rank one way on net present value and the opposite way on the criterion that actually answers the question.

So: the form decides who decides. Governance decides how well. The decision rules decide what gets built. Those three questions are the argument of everything below, and each of them ends in a number you have to be able to produce by hand.

Learning objectives

By the end you can:

1. Compare the organisational forms of both markets on the five dimensions that matter to an analyst (liability, taxation, capital access, continuity, disclosure), and compute the total tax burden of the same profit under an Indian LLP against an Indian private limited company, and under a US C corporation against a pass-through.
2. State why limited liability and the separation of ownership from control create the agency problem, and name which of the four principal conflicts a given fact pattern belongs to.
3. Compute the voting-versus-economic wedge on a dual-class or promoter-controlled register, and solve for the minimum economic stake consistent with retained voting control.
4. Map the governance mechanisms (board composition, the four standard committees, shareholder voting rights, covenants, regulation, the market for corporate control) to the specific conflict each is built to police, and say which mechanism is missing from a given company.
5. Read a board structure across three regimes: India under the Companies Act and SEBI's listing regulations, the United States under exchange listing standards, and the two-tier board of the German-speaking systems, including staggered boards and controlled-company exemptions.
6. Convert a governance assessment into a valuation adjustment by an explicit, capped rule, defend the size of the adjustment, show the identical result reached as a multiple haircut, and demonstrate that applying both routes at once double-counts.
7. Map ESG factors to the model line each one reaches, sector by sector, and reject the factors that reach no line.
8. Quantify ESG rating divergence by decomposing the gap between two providers into a weight effect, a score effect and an interaction that sum exactly to the gap, then restate the gap as a difference in percentile rank.
9. Turn an environmental exposure into arithmetic: a compliance shortfall into a margin, an abatement project into a net present value and a break-even carbon price, and a scope boundary into the difference it makes to both.
10. Complete the capital-budgeting toolkit: payback, discounted payback and the profitability index with each one's stated defect; mutually exclusive projects with unequal lives solved twice, by least common multiple and by equivalent annual annuity; capital rationing as a profitability-index ranking with the size of the error that ranking makes; and the incremental cash-flow principles that decide what enters the model at all.

Prerequisites & connections

Builds on. M0.05 taught you to find the Board's Report, the corporate governance report, the shareholding pattern and the DEF 14A proxy inside two real filings; here you learn to score them. M1.10 taught the disclosure furniture, including CARO 2020, the BRSR and the quarterly shareholding pattern under Regulation 31, and its 30-second CARO scan is the input to the checklist below. M3.01 owns present value, the annuity factor, net present value, the internal rate of return, the modified internal rate of return and the crossover rate, all of which are used here and none of which is re-derived. M3.07 owns capital allocation and the incentive question ("people do what they are paid to do"), which is the economic content behind every remuneration disclosure named below. EC1.02 owns share classes, differential voting rights, depositary receipts and index construction. E11.04 owns the Indian governance forensics: promoter-group mapping, the related-party-transaction web, Regulation 23 materiality tests, the pledge margin-call cascade and the auditor-signal sequence.

Feeds forward. The governance and ESG adjustments built here are consumed by M3.04 and M3.06 whenever a discount rate or a multiple is set for a controlled company. E11.01's credit lens uses the same evidence from the creditor's side, where the conflict is shareholder against lender rather than minority against controller. M8.06's document laboratory reads the primary sources at speed. The Phase 5 sector modules supply the materiality map in the ESG sections, because a factor is material only relative to a specific business model. And M10.02's India capstone is where a filled scorecard has to survive contact with a real company.

Four neighbouring nodes own material that touches this one, and none of it is rebuilt here. E11.04 owns the forensic depth on Indian governance: when a related-party web or a pledge cascade needs unpicking rather than merely scoring, that is where the work is done, and the checklist below deliberately stops at the point where a row is marked bad. EC1.02 owns everything about what a share class is, including the arithmetic of depositary-receipt parity and the mechanics of index inclusion; the wedge computed here uses that vocabulary without reteaching it. M3.01 owns the discounting machinery and the criteria a first course covers, so payback, the profitability index, unequal lives and capital rationing are added here as the completions and the crossover rate is left where it lives.

M1.10 owns the filing capstone and the BRSR walkthrough, so what follows treats the disclosure as an input and spends its words on what to do with it.

Core teaching content

8. Which environmental and social factors reach a line

A materiality map is a claim that a specific factor changes a specific line for a specific business model. Anything weaker belongs in the narrative section of a report where it can do no arithmetic harm.

Two rules keep a map like that from becoming a list. First, a factor belongs to a sector only if you can name the line and the direction; "reputational risk" names neither. Second, the same factor changes sign across sectors.

A carbon price is a cost to a cement maker and a revenue driver to a firm selling abatement equipment, and an analyst who scores both on the same environmental scale has thrown away the only information that mattered.

The arithmetic ledger

Every figure that carries an argument in the pages either side of this one, at the precision it was computed. If a number in a worked example, a solution or a key disagrees with this table, this table is right and the other is a transcription error worth reporting.

Common mistakes & how experts think differently

1. Treating a compliance box as a governance conclusion. A board can satisfy every numeric requirement in the listing rules and still be a rubber stamp, and a board that misses one requirement is not automatically captured. The expert reads composition for what the people can actually do: has anyone on the audit committee ever built a cash flow statement, has any independent director ever voted against anything, does attendance fall on the meetings where the difficult items sat. Compliance is a floor, and floors are where the least-governed companies stand.

1. Applying the governance view twice. A cost-of-equity add-on and a multiple haircut are the same statement in two currencies, and a note that applies both has taken roughly double the intended discount without saying so. Pick a door, state the size, show the other door's equivalent for the reader who prefers it, and never sum them.

1. Quoting a governance discount that cannot be switched off. If the adjustment is buried inside a forecast assumption, nobody can disagree with it in a useful way. The expert puts it in one cell, names the rule that produced it, and prints the unadjusted value beside the adjusted one, because a reader who rejects the opinion should still be able to use the model.

1. Confusing the wedge with the stake. A 16.36% economic holding that carries 66.18% of the votes is not a 16% problem. Compute the wedge and then compute the exit: how far can the controller sell and still control? At Northbridge the answer is down to 9.09%, and a controller who keeps full control on a 9% stake bears nine paise of every rupee of value it destroys.

1. Scoring environmental factors that reach no line. A sector-neutral score mixes a carbon price that is a cost for a cement maker with the same carbon price that is a revenue driver for an equipment supplier. The expert builds the materiality map first, keeps only the factors with a named line and a named direction, and treats everything else as narrative.

1. Reading a rating as a measurement. Two providers 1.95 points apart on a ten-point scale were 48.93 percentile points apart on the same company, and 1.50 points of the gap was disagreement about the pillar scores rather than about the weights, 1.25 of it in the environmental pillar alone. Decompose before you argue, and never compare ratings from two providers as though they were the same quantity measured twice.

1. Holding the margin flat while passing a cost through. Full pass-through of ₹65.3184 crore leaves EBITDA unchanged in rupees and still costs 0.2303 margin points, because revenue grew. A model that keeps the percentage margin constant has assumed a real improvement it never stated, and the error compounds through every year of the forecast.

1. Comparing unequal-lived projects on net present value. Machine B wins on net present value by $9,774.63 and loses on equivalent annual annuity by $5,173.71. The expert's reflex on seeing two mutually exclusive projects is to check the lives before reading either number, and to state the replication assumption when using either repair.

1. Ranking a rationed budget on the profitability index and stopping there. The ranking gave up ₹7.20 crore, 4.5627% of what the same budget could have produced, and left ₹50 crore unspent. Rank to build a shortlist, enumerate near the constraint, and always report the unspent balance.

1. Letting financing into the cash flow. Deducting interest and then discounting at a weighted average cost of capital charges for the debt twice and was worth -₹21.5644 crore at Saraswati, the largest of the five errors in that example and the hardest to see, because the resulting model is internally consistent and every subtotal foots.

1. Assuming the errors point the same way. Saraswati's five mistakes together produced a net present value of ₹21.3287 crore against a correct ₹24.2876 crore, a difference of only -₹2.9589 crore, because three errors worth -₹27.02 crore (the sunk study, the interest deduction and the wrong working-capital base) were almost cancelled by two worth +₹24.06 crore (the missing cannibalisation charge and the missing opportunity cost). A small total error is not evidence of a small mistake, and the expert audits each principle separately rather than sanity-checking the answer.

The thread through all eleven: an adjustment for governance or for an environmental factor is an opinion with a number attached, and its whole value comes from the reader being able to see where the opinion stops and the arithmetic starts. Print the rule, print the unadjusted case, print the sensitivity, and let the disagreement happen in the open.

Worked examples

Every figure below was computed in Python before it was written down, and the arithmetic ledger above restates each one at full precision. Every company named is synthetic. Regulatory thresholds carry a verify flag because they move.

Worked example 1: Scoring a governance checklist and pricing it (India, ₹)

Vindhya Ferro Alloys Limited (synthetic) is a listed Indian mid-cap with revenue of ₹3,050 crore, net profit of ₹186 crore, 38.0 crore shares outstanding and free cash flow to equity of ₹210 crore expected next year, growing at 6.00% in perpetuity. Its equity beta is 1.05, the risk-free rate is 7.00% and the equity risk premium is 6.50% (both illustrative; check against a current source before using them on a live name). Score its governance from the corporate governance report, the Board's Report and the shareholding pattern, then price the score.

Solution. Step 1, score ten rows, two points each. Each row is scored 2 where the practice is clean, 1 where it is compliant but weak, and 0 where it fails a rule or fails the analyst.

Total: 6 of 20, or 30.0000%. Row 1 is a rule failure, not a preference: with an executive promoter chairperson the listing regulations require at least half the board to be independent, which is 6 seats, so the company is 2 short (Regulation 17; check the current text). Row 5 sizes a channel at 13.5082% of revenue. Row 10 is the row that would send an analyst to the forensic work, which lives elsewhere in the curriculum and is not attempted here.

Step 2, convert the score by the stated rule. The house rule is a 3.00 percentage-point cap multiplied by the fraction of the checklist failed:

add-on = 0.0300 x (1 - 6/20) = 0.0300 x 0.70 = 2.1000 percentage points

Step 3, price it. The base cost of equity is 7.00% + 1.05 x 6.50% = 13.8250%, and the adjusted rate is 13.8250% + 2.1000% = 15.9250%.

The adjustment costs ₹567.8371 crore, or 21.1587% of the equity.

Step 4, verify by the other route. The implied price-to-earnings multiple falls from 14.4285 to 11.3756, which is a haircut of 21.1587%, identical to the value fall by construction. That identity is the check, and it is also the warning: an analyst who applies a 2.10-point add-on and then also marks the multiple down by the same proportion has taken 37.8405% off the equity for a single opinion.

Step 5, sensitivity, so the size can be argued with. The whole range the rule can produce is worth printing:

Read it. The 2.10 points are not a measurement of anything. They are a rule, stated in advance and applied to every company the same way, and the table is what makes the rule arguable. The most useful line in the whole example is the remediation arithmetic: separating the chair and appointing two independent directors takes the score from 6 to 10, cuts the add-on to 1.5000 points, and is worth ₹136.1417 crore, or ₹34.0354 crore per point. What the number cannot tell you is whether ₹412 crore of related-party transactions is ordinary trade or extraction, and no discount rate will ever answer that.

It sizes the exposure and hands the question to somebody willing to read the contracts.

Worked example 2: Two machines, different lives (US, $)

A US components plant must replace a press and has two mutually exclusive options at a 10.00% cost of capital. Machine A costs $420,000 and generates $135,000 of net cash flow a year for five years. Machine B costs $560,000 and generates $124,000 a year for eight years. Neither has a salvage value. Solve by least common multiple and by equivalent annual annuity, then compute payback, discounted payback and the profitability index for both.

Solution. Step 1, the annuity factors. AF(10%, 5) = 3.790786769 and AF(10%, 8) = 5.334926198.

Step 2, the raw comparison, which is the trap.

Machine B wins on net present value by $9,774.6347. That comparison is invalid, because it credits machine B with three years of service that machine A is not being asked to provide.

Step 3, the least common multiple. The lives are 5 and 8, so the common horizon is 40 years: eight replications of A and five of B, each chain discounted at 10.00%.

Machine A wins by $50,593.9520, reversing the raw ranking.

Step 4, the equivalent annual annuity. EAA = NPV / AF(r, n), so A returns 91,756.2139 / 3.790786769 = $24,205.0581 a year and B returns 101,530.8485 / 5.334926198 = $19,031.3502. Machine A wins by $5,173.7079 a year.

Step 5, verify that the two repairs are the same repair. AF(10%, 40) = 9.779050718. Multiply each annuity by it: A gives $236,702.4905 and B gives $186,108.5384, matching the chain values to the ninth decimal. The ratio of the chains, 1.271851858, equals the ratio of the annuities to nine decimals as well.

Step 6, the three completion criteria.

Read it. All three completion criteria pick machine A, which is the right answer, and none of them picked it for the right reason. A four-year payback cutoff would have rejected machine B without ever computing its net present value, and machine B is the project with the higher raw net present value; the rule got the decision right and the reasoning wrong, which is the worst kind of luck because it makes the rule look reliable. Machine A wins because it delivers more value per year of service, which is what the equivalent annual annuity measures and what payback cannot see.

One assumption is doing real work here, and it is replication at unchanged terms. For a press that is defensible. For a data centre it would not be, and the honest answer there is a shorter explicit horizon with a terminal value.

Worked example 3: A carbon compliance cost, through the margin and into the capital plan (India, ₹)

Kalinga Cement Works (synthetic) sells 8.4 million tonnes at ₹6,000 a tonne with a cash variable cost of ₹3,900 a tonne and ₹856.8000 crore of fixed cash cost, so revenue is ₹5,040.0000 crore and EBITDA is ₹907.2000 crore at an 18.0000% margin. Emissions intensity is 0.720000 tonnes of carbon dioxide per tonne of cement. A compliance mechanism requires a 6% intensity cut against that baseline, with the shortfall bought at an assumed ₹1,800 a tonne (an assumption, not a quote; check the current status of India's compliance carbon market before using any price). Quantify the margin effect, then evaluate an abatement project costing ₹340 crore that takes intensity to 0.660000, over fifteen years at 12.00%.

Solution. Step 1, the shortfall. Target intensity is 0.720000 x 0.94 = 0.676800, so the shortfall is 0.043200 tonnes per tonne of cement and 8,400,000 x 0.043200 = 362,880 tonnes in total. At ₹1,800 that is ₹65.3184 crore, which is 7.2000% of EBITDA and ₹77.76 per tonne of cement sold.

Step 2, three pricing responses.

Absorbing costs 1.2960 margin points. Passing through in full still costs 0.2303 points, because revenue rose by the full ₹65.3184 crore and EBITDA did not. Adding a demand response at an elasticity of -0.8 costs 1.0368% of volume and ₹18.2892 crore of EBITDA against the base.

Step 3, the abatement project. At an intensity of 0.660000 the company is 0.016800 tonnes per tonne below the target, so it holds 141,120 saleable credits worth ₹25.4016 crore a year. The annual swing is the ₹65.3184 crore of cost avoided plus the ₹25.4016 crore of income, or ₹90.7200 crore.

Step 4, find the assumption that carries it. The whole case rests on the carbon price, so solve for the price at which the fifteen-year net present value is zero. The annual swing is (362,880 + 141,120) x price, so 504,000 x price x 6.810864489 = ₹340 crore gives ₹990.4810 a tonne, 55.0267% of the assumed price. Recomputing the net present value at that price returns zero to nine decimals.

Read it. The number an investment committee needs is the last one. A project that still pays at ₹990 a tonne survives a 45% collapse in the carbon price, which converts an argument about climate policy into an argument about a threshold, and threshold arguments can be settled. Two things it does not show: it uses scope one and two emissions only, and it assumes the compliance target stays at a 6% cut rather than tightening, which would make the project better and the base case worse.

State both alongside the answer.

Worked example 4: Two providers, one issuer, and where the disagreement lives (US, $)

Cascade Grid Utilities Inc. (synthetic) is a US regulated utility rated by two providers on a ten-point scale. Provider P uses financial-materiality weights of 0.50 environmental, 0.20 social and 0.30 governance, and scores the pillars 3.0, 6.0 and 8.0. Provider Q uses 0.25, 0.35 and 0.40, and scores 5.5, 6.5 and 8.5. Decompose the gap, then restate it as a difference in rank against a universe with mean 5.60 and standard deviation 1.40.

Solution. Step 1, the totals. P = 0.50(3.0) + 0.20(6.0) + 0.30(8.0) = 5.1000. Q = 0.25(5.5) + 0.35(6.5) + 0.40(8.5) = 7.0500. The gap is 1.9500 points.

Step 2, decompose it pillar by pillar.

The three effects sum to the gap with a residual of zero to twelve decimal places, which is the identity check and the reason to run the decomposition at all.

Step 3, read where the disagreement is. The score effect is 1.5000 of the 1.9500 gap, 76.9231% of it, and 1.2500 of that 1.5000, or 83.3333%, is the environmental pillar alone. The two providers do not mainly disagree about how much the environment matters for a utility; they disagree about this utility's environmental performance, almost certainly because one includes financed or supply-chain emissions in a boundary the other excludes. The weight effect of +0.9500 is real but is partly cancelled by an interaction of -0.5000, which arises because provider Q cut the weight on exactly the pillar where it scores the company higher.

Step 4, convert the gap to a rank. On the stated universe distribution, 5.1000 sits at the 36.0492nd percentile and 7.0500 at the 84.9832nd. The 1.9500-point gap is 1.3929 standard deviations and 48.9340 percentile points. An exclusion screen keeping the top 45% of the universe, which cuts at 5.7759, drops the company on provider P's data and keeps it on provider Q's.

Read it. A rating difference that looks like a rounding error on a ten-point scale is the difference between owning and not owning the stock. And the decomposition tells you what to do about it: since the disagreement is a score disagreement inside one pillar, the resolution is to read both providers' environmental methodology documents and decide whose boundary you accept. If the gap had been a weight effect, no amount of methodology reading would help, because the two providers would then be answering different questions and you would have to pick the question first.

The universe mean and standard deviation are assumptions here. With a real universe, use the empirical distribution, because these scores are usually not normal.

Worked example 5: Six projects, one budget (India, ₹)

A diversified Indian manufacturer has six independent, indivisible projects and a hard capital budget of ₹500 crore for the year, at a 12.00% cost of capital. Rank them, choose a bundle, and quantify what the ranking costs.

Solution. Step 1, the table.

All six have a profitability index above 1.0000, so all six create value; taking every one would need ₹820 crore against a ₹500 crore budget and would produce ₹257.10 crore of net present value.

Step 2, rank by profitability index and fill the budget. The order is C (1.3800), A (1.3500), E (1.3200), D (1.3000), B (1.2700), F (1.2200). Take C at ₹120 crore, A at ₹180 crore and E at ₹90 crore, reaching ₹390 crore; D at ₹220 crore and B at ₹150 crore both overflow; F at ₹60 crore fits. The bundle is C + A + E + F, spending ₹450 crore for ₹150.60 crore of net present value, with ₹50 crore unspent.

Step 3, enumerate every feasible bundle. With six projects there are 63 non-empty subsets and the search is instant.

The best feasible bundle is A + D + E at ₹490 crore for ₹157.80 crore. Ranking by profitability index gave up ₹7.20 crore, which is 4.5627% of what the budget could have produced.

Step 4, the upper bound. Allow a fractional project and the ranking becomes optimal by construction: C, A and E fill ₹390 crore, and the remaining ₹110 crore buys 50.00% of D, adding ₹33.00 crore for a total of ₹170.40 crore. The three answers form a ladder of ₹150.60, ₹157.80 and ₹170.40 crore, and the ₹12.60 crore gap between the integer optimum and the bound is the pure cost of indivisibility.

Read it. The profitability index is the right ranking device and the wrong stopping rule, because a high-index project can crowd out a pair that together use the budget better. Honesty requires one more line: ranking by net present value instead happens to find A + D + E here, so on this data the cruder heuristic wins. That is luck rather than a result, and constructing a case with one very large project makes net-present-value ranking fail just as badly.

Rank for a shortlist, enumerate near the constraint, and report the unspent ₹50 crore, because the unspent balance is the price of the budget and is the number most likely to get the budget changed.

Worked example 6: A controlled company, priced twice (US, $)

Northbridge Media Corp. (synthetic) has 46.0 million Class A shares at one vote and 9.0 million Class B shares at ten votes, all Class B held by the founder group, with Class B converting one-for-one into Class A on sale. EBITDA is $412 million, net debt is $780 million, and comparable single-class peers trade at 11.0 times EBITDA. Free cash flow to equity is $196 million growing at 3.00%. Compute the wedge and the control floor, then price the structure as a multiple haircut and check it as a discount-rate add-on.

Solution. Step 1, the register. Total shares are 55.0 million and total votes are 46.0 + 9.0 x 10 = 136.0 million. The founder group holds 9.0 / 55.0 = 16.3636% of the economics and 90.0 / 136.0 = 66.1765% of the votes, a wedge of 49.8128 points and a ratio of 4.0441 times.

Step 2, the control floor. Selling s million Class B shares leaves the group 10(9 - s) votes out of 136 - 9s. Setting the ratio to one half:

90 - 10s = 0.5(136 - 9s) -> 90 - 10s = 68 - 4.5s -> s = 4.0

The group can sell 4.0 million shares, 44.4444% of its holding, fall to a 9.0909% economic stake, and still hold exactly half the vote.

Step 3, price it as a multiple haircut. The company qualifies for the controlled-company exemption from majority board independence and from independent compensation and nominating committees (check the current exchange listing standards), so a haircut is defensible. Apply 12%:

Step 4, notice the amplification. Enterprise value fell 12.0000% and equity fell 14.4947%, a factor of 1.2079, because net debt is fixed at 17.2109% of the unadjusted enterprise value. A haircut applied to an enterprise-value multiple always lands harder on the equity, and the more levered the company the harder it lands.

Step 5, cross-check through the discount rate. Treating the equity as a growing perpetuity, Re = 196 / 3,752 + 0.03 = 8.2239% unadjusted and 196 / 3,208.16 + 0.03 = 9.1094% adjusted. The 12% multiple haircut is 88.55 basis points of cost of equity for this company at these numbers.

Read it. The two routes agree because they are the same statement, which is the whole reason to run both: the agreement is the check, and it makes the double-counting error impossible to commit accidentally. What the arithmetic will not tell you is whether 12% is the right haircut. That comes from the checklist, from whether the dual-class structure has a sunset provision, and from what the controller has actually done with the votes over the past five years.

A founder group at a 9.09% economic stake with permanent control is a different security from one at 40%. Neither the multiple nor the discount rate knows the difference until you tell it.

Worked example 7: What belongs in the cash flow (India, ₹)

Saraswati Snack Foods (synthetic) is evaluating a baked-snacks line. A market study costing ₹4.20 crore has already been paid for. The line costs ₹96.0 crore, is depreciated straight-line over eight years to zero, and generates ₹120.0 crore of revenue against ₹84.0 crore of cash operating cost. It will take ₹18.0 crore of revenue from the existing fried-snacks line, which runs at a 22% EBITDA margin. The site is idle but leasable at ₹2.40 crore a year. Working capital runs at 12% of incremental revenue, invested at the start and recovered at the end. A ₹60 crore term loan at 9.50% funds part of it, costing ₹5.70 crore of interest a year. The tax rate is 25.168% and the weighted average cost of capital is 11.50% (rates illustrative; check the current corporate rate before using it).

Solution. Step 1, build the incremental EBITDA.

Step 2, to operating cash flow. Depreciation is 96.0 / 8 = 12.0000, so incremental EBIT is 29.6400 - 12.0000 = 17.6400. Tax at 25.168% is 4.4396, NOPAT is 13.2004, and adding depreciation back gives ₹25.2004 crore a year.

Step 3, working capital. Incremental revenue is 120.0 - 18.0 = 102.0, so working capital is 0.12 x 102.0 = ₹12.2400 crore at time zero, recovered at year eight.

Step 4, the net present value. The stream is -108.2400 at time zero, +25.2004 in years one to eight, and +12.2400 more in year eight. With AF(11.5%, 8) = 5.055636778, the value is ₹24.2876 crore and the internal rate of return is 17.3566%.

Step 5, run the wrong version and attribute the error. Take a model that capitalises the study, ignores cannibalisation and the opportunity cost, deducts interest, and applies working capital to the full ₹120 crore. It gives an annual cash flow of ₹25.6943 crore on an outlay of ₹114.6000 crore, for a net present value of ₹21.3287 crore. Each error taken alone, from the correct base:

Read it. The five effects sum to exactly the joint effect here because each one shifts a level rather than interacting, and the total is a mere -₹2.9589 crore, 12.1827% of the correct answer. That small total is the trap. Two errors worth a combined +₹24.06 crore were almost exactly offset by three worth -₹27.02 crore, and an analyst who checked only the final number would have concluded the model was roughly right. The largest single error is the interest deduction, at -₹21.5644 crore, and it is also the least visible, because a model that subtracts interest and then discounts at a weighted average cost of capital foots perfectly at every subtotal.

Audit the principles one at a time, not the answer.

Practice set

Work each problem before reading its solution. A calculator is assumed and full precision is expected in the working; round only at the final figure, to the precision the problem asks for. Where a tolerance is not stated, treat an answer within 0.5% of the stated one as correct. Problems alternate between the two markets on purpose.

P1 (guided). The form decides the tax bill (India, ₹). A professional services business earns ₹9.00 crore of profit and distributes all of it. Compare an LLP taxed at 30% plus a 12% surcharge plus a 4% cess against a private limited company taxed at 22% plus a 10% surcharge plus a 4% cess, with the dividend then taxed in the recipient's hands at an effective 31.2%. Compute both effective rates, then find the payout ratio at which the two forms leave the owners with identical cash.

Solution. LLP: 0.30 x 1.12 x 1.04 = 34.9440%, so tax is ₹3.1450 crore and the owners keep ₹5.8550 crore. Company: 0.22 x 1.10 x 1.04 = 25.1680%, so entity tax is ₹2.2651 crore and ₹6.7349 crore is distributable; dividend tax of ₹2.1013 crore takes total tax to ₹4.3664 crore, an effective 48.5156%, leaving ₹4.6336 crore. The gap is 13.5716 points in favour of the LLP on full distribution. Break-even payout: solve 6.7349 - p x 6.7349 x 0.312 = 5.8550 to get p = 41.8716%. Below that payout the company form wins, because the second layer only fires on what is distributed. Verify all four rates against the current Finance Act before using them.

P2 (guided). The wedge, and how far it stretches (US, $). A company has 120 million Class A shares at one vote and 30 million Class B at ten votes, all Class B held by the founders, converting one-for-one on sale. Compute the economic stake, the voting stake and the wedge, then find the lowest economic stake consistent with holding at least half the votes.

Solution. Economic stake 30 / 150 = 20.0000%. Votes 300 / (120 + 300) = 71.4286%. Wedge 51.4286 points. Selling s leaves 10(30 - s) votes out of 420 - 9s; setting the ratio to one half gives 300 - 10s = 210 - 4.5s, so s = 16.3636 million, and the founders hold 13.6364 / 150 = 9.0909% of the economics with exactly half the vote. The 9.0909% floor is the same as Northbridge's, which is not a coincidence: with a ten-to-one vote ratio and one-for-one conversion, the floor depends only on the vote multiple.

P3 (guided). Counting seats (India). A listed company has a board of 12 with an executive chairperson who is a member of the promoter group, of whom 5 are independent. Its audit committee has 3 members, 2 of them independent, chaired by an independent director. State the compliance position on both, and give the independent share of the board.

Solution. Independent share is 5 / 12 = 41.6667%. With an executive promoter chairperson the listing regulations require at least half the board independent, which is 6 of 12, so the company is 1 director short (Regulation 17). The audit committee needs at least three directors with two thirds independent, which is ceil(2/3 x 3) = 2, and an independent chair; with 2 of 3 independent and an independent chair it is compliant (Regulation 18). Note the asymmetry the arithmetic creates: adding a fourth member would raise the independence requirement to ceil(2/3 x 4) = 3, so a committee can fall out of compliance by growing.

P4 (guided). Three criteria on one project (US, $). A project costs $250,000 and returns $90,000 a year for four years at a 9.00% cost of capital. Compute payback, discounted payback, net present value and the profitability index, and say which criteria a three-year payback policy and a "profitability index above 1.20" policy would each reject.

Solution. Payback 250,000 / 90,000 = 2.7778 years. Discounted flows are 82,568.81, 75,751.20, 69,496.51 and 63,758.27, cumulating to 82,568.81, 158,320.01, 227,816.52 and 291,574.79, so discounted payback is 3 + (250,000 - 227,816.52) / 63,758.27 = 3.3479 years. With AF(9%, 4) = 3.239719877, the present value of inflows is $291,574.79, the net present value is $41,574.79 and the profitability index is 1.1663; the internal rate of return is 16.3675%. A three-year payback policy accepts it on undiscounted payback and rejects it on discounted payback, which is a good demonstration that the two rules are different policies rather than a rough and a precise version of one. A "profitability index above 1.20" hurdle also rejects it, despite a positive net present value, which is what an arbitrary index hurdle does.

P5. Equivalent annual cost (India, ₹). Two effluent treatment plants: X costs ₹34.0 crore with ₹6.4 crore of annual operating cost over a 6-year life; Y costs ₹52.0 crore with ₹5.1 crore of annual cost over a 10-year life. The discount rate is 10.00%. Choose, and verify by replication.

Solution. AF(10%, 6) = 4.355260699 and AF(10%, 10) = 6.144567106. Present value of cost: X is 34.0 + 6.4 x 4.355260699 = ₹61.8737 crore; Y is 52.0 + 5.1 x 6.144567106 = ₹83.3373 crore. Equivalent annual cost: X is ₹14.2067 crore a year, Y is ₹13.5628 crore a year, so Y wins by ₹0.6439 crore a year despite costing 34.6894% more in present value. Verify over the least common multiple of 30 years: X's chain costs ₹133.9249 crore and Y's ₹127.8550 crore, and each equals its annual cost times AF(10%, 30) to nine decimals. Choosing on present value of cost alone would have picked X and been wrong by ₹6.07 crore over thirty years.

P6. Rationing, again (US, $). Five indivisible projects and a $1,000,000 budget. P: outlay $260,000, present value of inflows $358,800. Q: $240,000 and $326,400. R: $220,000 and $294,800. S: $640,000 and $851,200. T: $300,000 and $393,000. Rank by profitability index, take the greedy bundle, find the integer optimum and the divisible upper bound.

Solution. Profitability indices: P 1.3800, Q 1.3600, R 1.3400, S 1.3300, T 1.3100. Net present values: P $98,800, Q $86,400, R $74,800, S $211,200, T $93,000. Greedy: P then Q then R fills $720,000 for $260,000, and neither S nor T fits in the remaining $280,000. The integer optimum is P + S at $900,000 for $310,000, so the ranking gave up $50,000, or 16.1290%. The divisible bound takes P, Q, R and 43.75% of S for $352,400. This is the crowd-out case: S has the fourth-best index and the best net present value, and any rule that reaches it last will never fit it in.

P7. A compliance shortfall (India, ₹). A synthetic steel maker ships 4.2 million tonnes at an intensity of 2.30 tonnes of carbon dioxide per tonne, against a target of 2.18, with credits at ₹2,200 a tonne. Revenue is ₹8,600 crore and EBITDA ₹1,420 crore. Compute the compliance cost, the effect on EBITDA and margin, and the price rise needed for full pass-through on a ₹52,000 a tonne realisation.

Solution. Emissions 4,200,000 x 2.30 = 9,660,000 tonnes against a target of 9,156,000, a shortfall of 504,000 tonnes. Cost 504,000 x 2,200 = ₹110.8800 crore, which is 7.8085% of EBITDA. Margin falls from 16.5116% to 15.2223%, a fall of 1.2893 points. Full pass-through needs 110.8800 crore / 4,200,000 t = ₹264.00 a tonne, which is 0.5077% of a ₹52,000 realisation. That last figure is the one to lead with in a note: a cost worth nearly 8% of EBITDA is worth half a percent of price, so the question is entirely about whether the industry can hold half a percent, not about whether the cost is large.

P8. Where two raters disagree (US). Provider P weights the pillars 0.40, 0.25 and 0.35 and scores them 4.5, 5.0 and 6.0. Provider Q weights them 0.30, 0.30 and 0.40 and scores them 6.0, 4.0 and 6.5. Decompose the gap and verify the identity.

Solution. P = 1.80 + 1.25 + 2.10 = 5.1500 and Q = 1.80 + 1.20 + 2.60 = 5.6000, so the gap is 0.4500. Weight effect (-0.10)(4.5) + (0.05)(5.0) + (0.05)(6.0) = +0.1000. Score effect 0.40(1.5) + 0.25(-1.0) + 0.35(0.5) = +0.5250. Interaction (-0.10)(1.5) + (0.05)(-1.0) + (0.05)(0.5) = -0.1750. The three sum to 0.4500 with a residual of zero to twelve decimals. The two providers disagree sharply on two pillars in opposite directions, +1.5 on the environmental and -1.0 on the social, and the aggregate gap of 0.45 hides both of those disagreements. A near-identical total is not agreement.

P9. Score to value, second issuer (India, ₹). A company scores 13 of 20 on the same checklist, has a beta of 1.15, and generates ₹96 crore of free cash flow to equity growing at 5.00%, with a 7.00% risk-free rate and a 6.50% equity risk premium. Using the 3.00 percentage-point cap rule, compute the add-on and the value effect.

Solution. Add-on 0.0300 x (1 - 13/20) = 1.0500 percentage points. Base cost of equity 7.00% + 1.15 x 6.50% = 14.4750%, adjusted 15.5250%. Value 96 / (0.144750 - 0.05) = ₹1,013.1926 crore unadjusted and 96 / (0.155250 - 0.05) = ₹912.1140 crore adjusted, a fall of 9.9762%. Compare with Vindhya's 21.1587% on a 6-of-20 score. Half the failure rate gives exactly half the add-on, but the value effect is 47.1496% of Vindhya's rather than 50%, because this issuer's denominator Re - g is 9.4750 points wide against Vindhya's 7.8250, and a wider denominator absorbs a given add-on with less damage. Never assume the value effect scales with the score.

P10. What goes in the model (US, $). A project generates $4,200,000 of revenue against $2,520,000 of variable cost. It cannibalises $310,000 of contribution from an existing product. It occupies a warehouse the firm could sublet for $145,000 a year. Head office allocates $260,000 of existing overhead to it, none of it new. Depreciation is $900,000 and interest on project-linked debt is $168,000. The tax rate is 21%. Compute the incremental after-tax operating cash flow, and the figure a model that used the allocation and deducted interest would have produced.

Solution. Incremental EBITDA 4,200,000 - 2,520,000 - 310,000 - 145,000 = $1,225,000. EBIT 1,225,000 - 900,000 = $325,000; tax $68,250; after-tax operating cash flow 256,750 + 900,000 = **$1,156,750**. The mistaken version substitutes the $260,000 allocation for the two real charges and deducts $168,000 of interest, giving (4,200,000 - 2,520,000 - 260,000 - 900,000 - 168,000) x 0.79 + 900,000 = $1,178,080, overstating by $21,330. The allocated overhead is the item most often defended in a meeting, and it is the easiest to refuse: if the cost exists whether or not the project proceeds, it is not incremental, whatever the cost accounting says.

P11. Two American forms (US, $). A business earns $2,000,000. As a C corporation it pays 21% at the entity and the shareholder pays 23.8% on the dividend. As a pass-through the owner pays 37%, or 37% on 80% of the income if the qualified business income deduction applies. Compare the effective rates and find the retention rate at which the C corporation catches up.

Solution. C corporation: entity tax $420,000, dividend tax 1,580,000 x 0.238 = $376,040, total $796,040, an effective 39.8020%. Pass-through without the deduction: $740,000, or 37.0000%. With the 20% deduction: 2,000,000 x 0.80 x 0.37 = $592,000, or 29.6000%, a gap of 10.2020 points against the C corporation. Solving for the payout at which the C corporation route leaves the owner the same after-tax value as the deduction case, treating retained earnings as worth their face value, gives 45.7398%, so full retention favours the corporation and full distribution does not. Every rate here carries a verify flag, and the deduction in particular has a statutory sunset that has been amended more than once; check the current code before quoting any of it.

P12. Lives and governance together (India, ₹). Two mutually exclusive plant options at a 12.50% cost of capital. Option A costs ₹62.0 crore and returns ₹21.0 crore a year for 5 years. Option B costs ₹88.0 crore and returns ₹19.6 crore a year for 9 years. Rank them on net present value and on equivalent annual annuity, then re-rank after a governance review adds 1.50 percentage points to the discount rate.

Solution. AF(12.5%, 5) = 3.560568342 and AF(12.5%, 9) = 5.228484671. Net present values: A 21.0 x 3.560568342 - 62.0 = ₹12.7719 crore; B 19.6 x 5.228484671 - 88.0 = ₹14.4783 crore, so B wins on net present value. Equivalent annual annuity: A 12.7719 / 3.560568342 = ₹3.5870 crore a year; B 14.4783 / 5.228484671 = ₹2.7691 crore, so A wins on the criterion that answers the question. At 14.00% the factors become 3.433081 and 4.946372, giving net present values of ₹10.0947 crore and ₹8.9489 crore and annuities of ₹2.9404 crore and ₹1.8092 crore, so A now wins on both. The governance add-on did not change the right answer; it changed the wrong answer into the right one, which is a reason to compute the annuity in the first place rather than to trust the add-on.

Applied mini-project

Build a governance and environmental scorecard that can be switched off.

Pick one listed company you can get primary documents for, in either market. Indian names give you the corporate governance report, the Board's Report annexures, the shareholding pattern with the pledge table and the BRSR in one document; American names give you the 10-K and the DEF 14A. Build four artefacts, each of which must pass a stated test before it counts toward the score.

Artefact 1 is the filled checklist, and it carries two tests. Score the ten rows from Worked Example 1 against the primary documents, and for each row record the exact page or table you took the fact from and the sentence in the valuation that the row changes. Test 1: every row cites a page in a document you can produce, and no row is scored from a summary, a database or a rating. Test 2: the two rows that are rule tests rather than judgement calls, board independence and audit committee composition, are checked against the current text of the applicable regulation, quoted with its number and the date you read it.

A row you cannot source scores zero and is reported as NOT DISCLOSED rather than guessed.

Artefact 2 is the materiality map, with one test. For the company's sector, list the environmental and social factors that reach a line in your model, naming the line and the direction for each, and list the factors you considered and rejected with one sentence saying why each reaches no line. Test 3: every retained factor has a line and a sign, and the rejected list is not empty. A map with nothing rejected has not been built, only copied.

Artefact 3 is the separable valuation, and it carries three tests. Produce a base valuation with no governance or environmental adjustment, then a schedule of adjustments in which each one is a single named input. Apply the discount-rate route and show the multiple-haircut equivalent for the same adjustment. Test 4: setting every adjustment cell to zero reproduces the base valuation to the last rupee or cent. Test 5: the discount-rate route and the multiple route agree on the adjusted value, and you state which one you are reporting. Test 6: a sensitivity table shows the value at the adjustment, at half of it, and at the cap, so a reader can locate their own view.

Artefact 4 is one quantified environmental exposure, with one test. Take the single most material factor from Artefact 2 and carry it all the way to a number: a compliance cost or a physical exposure converted into a margin or a capital-expenditure line. Evaluate the abatement or mitigation option as a capital-budgeting decision, using at least the net present value, the equivalent annual annuity where lives differ, and the profitability index. Test 7: solve for the assumption break-even, meaning the value of the driving input at which the exposure stops mattering, and state where the current input sits against it.

What you hand in is one workbook or notebook plus a memo of at most three pages. The memo carries the filled checklist with its sources, the materiality map, the base and adjusted valuations side by side, and the seven test results.

Close it with the two facts that would most change your answer, and what you would have to read to settle them.

Rubric (21 points; pass at 17, with no zero on any row).

Any artefact whose test fails scores zero for that row regardless of how well the memo reads. A valuation whose adjustments cannot be switched off is not a valuation with a governance view in it; it is a valuation nobody else can use, and that is the specific failure this rubric exists to prevent.

Reading & resources

Each entry here is either a primary source to verify against or a framework to learn from, and none of them is a source to quote a number from without tracing it yourself.

Primary law and rules, which move and must be checked on the day you use them:

• SEBI Listing Obligations and Disclosure Requirements Regulations, 2015 (current amended version at sebi.gov.in): Regulation 17 on board composition, 18 on the audit committee, 19 on nomination and remuneration, 23 on related-party transactions, 31 on the shareholding pattern. The single most useful document for Indian governance work. [Free] [Intermediate]
• The Companies Act, 2013 (mca.gov.in): sections 134 and 135 on the Board's Report and corporate social responsibility, 139 on auditor appointment and rotation, 149 on independent directors, 177 and 178 on committees, 188 on related-party contracts. [Free] [Intermediate]
• SEBI's BRSR framework and the BRSR Core circulars (sebi.gov.in): the format, the applicability threshold and the assurance glide path, all of which have been revised. [Free] [Beginner]
• NYSE Listed Company Manual section 303A and the Nasdaq 5600 rules: the American board and committee floors, and the controlled-company exemption that removes most of them. [Free] [Intermediate]
• The UK Corporate Governance Code (frc.org.uk) and the OECD Principles of Corporate Governance: the comply-or-explain design and the international reference text against which most national codes are written. [Free] [Beginner to Intermediate]

Frameworks and the academic spine:

• Adolf A. Berle and Gardiner C. Means, The Modern Corporation and Private Property (1932): where the separation of ownership from control was first laid out at length. [Paid] [Advanced]
• Michael C. Jensen and William H. Meckling, "Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure", Journal of Financial Economics (1976): the monitoring, bonding and residual-loss decomposition used throughout. [Free] [Advanced]
• Andrei Shleifer and Robert W. Vishny, "A Survey of Corporate Governance", Journal of Finance (1997): the best single map of the field's questions. [Paid] [Advanced]
• Paul Gompers, Joy Ishii and Andrew Metrick, "Corporate Governance and Equity Prices", Quarterly Journal of Economics (2003), and Lucian Bebchuk, Alma Cohen and Allen Ferrell, "What Matters in Corporate Governance?", Review of Financial Studies (2009): the governance index and the six-provision entrenchment index, and the argument between them about which provisions carry the effect. [Paid] [Advanced]

Sustainability reporting and materiality:

• The ISSB standards (IFRS S1 and S2) and the SASB Standards they absorbed: the sector-by-sector materiality maps that make the environmental and social work into arithmetic rather than narrative. Read your sector's standard, not the general one. [Free] [Intermediate]
• The GHG Protocol Corporate Standard: what scopes one, two and three actually mean, and why a boundary choice moves a headline number more than any operating change. [Free] [Beginner]
• The EU Corporate Sustainability Reporting Directive and the ESRS, plus the amendments that have rescoped both: the broadest mandatory regime, and the one whose timetable has changed most. [Free] [Intermediate]

Capital budgeting:

• Richard A. Brealey, Stewart C. Myers and Franklin Allen, Principles of Corporate Finance: the standard treatment of payback, the profitability index, unequal lives and rationing, with the equivalent annual cost worked properly. [Paid] [Beginner to Intermediate]
• Tim Koller, Marc Goedhart and David Wessels, Valuation: Measuring and Managing the Value of Companies: the incremental cash-flow discipline and the reason financing never enters the flow. [Paid] [Intermediate]

One exercise is worth more than any of the reading above. Take a company you already follow, open its most recent governance report or proxy, and time yourself filling the ten-row checklist. Most people take forty minutes the first time and twelve the third. Then do the same company a year earlier and compare the two: the rows that moved, and the rows that were quietly redefined between the two documents, are where the year's real governance news is.

Teach it back & journal

The Feynman half. Take a colleague who values companies well and has never scored a proxy. Put the Northbridge register in front of them, 46 million shares at one vote and 9 million at ten, and ask them for the founders' economic stake. They will say 16.36% and be right. Then ask how far the founders could sell and still control the company, and let them work it out to 9.09% themselves, because the number lands differently when you derive it.

Now hand them the two doors. Show them the 12% multiple haircut and the 88.55 basis-point add-on producing the same $58.33 a share, and make them say out loud why adding both would be wrong. Finish with the amplification: ask why a 12% cut to enterprise value became a 14.49% cut to equity, and do not accept an answer that does not contain the words "net debt did not move".

Then the part people resist. Give them the two machines and ask which is better. Almost everyone picks machine B, because $101,531 is larger than $91,756. Let them commit to it in writing before you show them the forty-year chain. Making them commit is the whole device, because the error is not a knowledge gap; they know about equivalent annual annuities. The question simply did not look like the kind of question that needed one, and the only durable fix is the reflex of checking the lives before reading either number.

The journal half. For the next two weeks, catch every governance or sustainability claim you meet in a published note, a rating, a term sheet or a company presentation. Write down three things before you use any of them: which of the four conflicts the claim is about, which line in a model it changes and in which direction, and what the person making the claim would have had to read to know it. Then answer in writing at the end. Count the claims that survived all three tests, the ratings that named a weighting scheme, and the carbon figures that named a scope boundary. Where a claim reached no line at all, write down what you did with it and what you should have done.

Then write a paragraph on the harder thing. The adjustments built here are opinions with numbers attached, and the temptation runs in two directions at once: to leave the opinion out because it cannot be measured, or to bury it inside a forecast where nobody can argue with it. Which of the two are you more likely to do under time pressure? Write down what specifically goes wrong when you do it, and one concrete change to how you lay out a model that would make the other choice easier than the one you default to.

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.