The Agency Problem: When Managers and Owners Diverge

Two gaps open between the people who run a business and the people whose business it is: they want different things, and one side knows more. Neither gap can be seen from outside. The one thing a reader outside can see is the list of who the business is connected to, and at this one that list runs to four lines with two of them reading none listed.

Where does this problem actually sit, and is it an accusation against anybody?

Part of the answer is settled where corporate governance is the subject in its own right. The governance treatment puts it in one sentence: whoever controls the money day to day can serve themselves before serving the owners, and that is not an accusation against anyone, it is a description of where the opportunity sits.

Almost everybody skips the second half of that sentence, so read it again. The claim is about a position, not about a person standing in it. The claim is true of a business run beautifully for forty years by somebody who never took a rupee that was not theirs. The claim is true of the shop at the end of any street. The position is the reason governance structures exist at all, in the same way that a lock on a cash drawer is a statement about drawers rather than a statement about the person holding the key.

Every household already runs this arrangement. A household hands the week's shopping money to whichever member is going to the market. Nobody in that household suspects anybody. One person is simply now between the money and everybody else, and only that person knows what the vegetables actually cost that morning. Nothing has gone wrong. Something has become possible, and everyone involved would say the same.

The same treatment names the two parts of the space that opens up, and names them in a way worth keeping. The first is a gap of interest. The runner would like a bigger salary, a bigger empire, a quieter life, or a good price for a relative. The owner would like the money used well. The second is a gap of information: the runner sees every invoice and every offer, and the owner sees a report four times a year, written by the runner. The governance treatment works that pair out at length on Aravalli Agro Foods, a listed snack maker invented for it, and reaches a conclusion carried across here without reopening: every governance structure is a variation on one move. Reduce the gap of interest, or reduce the gap of information, or both, at the point where money is about to move.

The single move is the whole of the mechanism. Committees, approval rights, independent directors, disclosure duties: each one is that move wearing a different set of clothes, and each is covered separately. A much narrower and much more awkward question is left over. Granted that the two gaps exist everywhere, what can somebody standing outside a business actually see of them?

So why can the gap not simply be looked for?

The first thing to establish is bleak. Take the two gaps one at a time and ask what instrument would measure each: the honest answer both times is that no instrument exists. Not because nobody has built one yet. Because of what each gap is made of.

The gap of interest is a fact about what two people want. Wanting is not a line item. In any set of accounts at all, down every statement and every note, there is no row headed what the manager would have preferred. There never will be. The accounts record what a business did with money, and a preference that was never acted on moved no money at all. Even a preference that was acted on arrives in the accounts wearing the same clothes as every other decision: a cost went up, a purchase happened, a project was approved. The wanting stays behind the number and does not come out.

The gap of information is worse, and it is worse in a way that is almost funny once it is seen. The information gap is defined by everything the reader was never given. Any attempt to measure it has to use the published record, and the published record is precisely the side of the gap the reader is standing on. Measuring the size of what somebody else knows, using only the things that somebody disclosed, is not a hard problem. The measurement is a problem with no shape.

A gap that cannot be measured directly is not therefore unknowable, and it is also not therefore absent. The two halves of that sentence are doing different jobs and both matter. The first half keeps the work going: an indirect route exists, and everything below is that route. The second half is the one that saves careers.

The failure it saves a reader from is not a rare one. A reader sets out to find evidence of a divergence, looks carefully, finds nothing, and writes down that there is none. The sequence repays slow reading. The search returned nothing because the search was never capable of returning anything else. Its output is the same whether a gap is wide open and acting or shut tight and dormant, so the finding carries no information at all, and the reader has just converted a method's blind spot into a conclusion about a business.

The market stall has the same shape. Somebody is sent to buy vegetables and comes back with vegetables and the right change. The return establishes that they came back with vegetables and the right change. The return establishes nothing whatsoever about the price they were quoted, the price they paid, or whether the seller is their cousin. None of those things was ever going to travel home in the bag.

So the useful question is a different one, and it is worth writing down before anything else arrives. Where is the gap cannot be answered. Instead: which of this business's decisions would have left a trace if a gap had been acting, and which would have left none? The question has an answer, the answer is short, and getting it is the whole of the skill.

What does this business's own list of connections say, before any amount is attached?

There is exactly one route a divergence can travel that the accounts were built to catch, and it is worth stating precisely because everything after this depends on it. If interests diverge and money moves, the money moves somewhere. If it moves toward somebody connected to the people deciding, a disclosure duty attaches to it and a note in the accounts sets it out. The note is free to read and takes about twenty seconds.

Work it on a real published list. Anjani Stationers Private Limited is a stationer invented for these notes and trading nowhere, making registers for schools. Its note of related partyA person or another business close enough to the reporting business that ordinary bargaining between strangers cannot be assumed to have settled their dealings. Companies inside the same group and the senior people who direct the business are the usual cases. disclosures is set out in full in these notes, and the whole thing is four lines.

The note groups connections by relationship rather than by size. The grouping is itself the first reading. The usual groupings are the companies inside the same group, the people who direct the business, the other entities those people control, and the close relations of those people. Each grouping is a different kind of closeness. For Anjani Stationers the four lines read like this. Companies inside the same group: Chitra Binding Works, 70 per cent held from April. The people who direct the business: Vaidehi Rao, finance controller. Other entities those people control: none listed. Close relations of those people: none listed. The working that publishes it calls that the whole surround.

The list alone already says a great deal, with not one rupee attached to any of it. Anjani Stationers has a binding operation inside its own group from April of its second year, and had none in the first year. Chitra Binding Works Private Limited is a subsidiaryA company in which another company holds a large enough share to decide how it is run. The larger company does not have to hold all of it, and usually does not. from that month, and binding is a thing this business used to buy from outside. A subsidiary in the same trade changes what the rest of the accounts are expected to look like before another line of them has been read, and it settles which questions are worth carrying forward.

The list also says what kind of business this is. One entry among the key management personnelThe people with authority and responsibility for planning, directing and controlling the activities of a business, directly or indirectly. Directors and the most senior managers are the usual members of this grouping., and it is a single named finance controllerThe senior person in charge of a business's books and its reporting. It is a job somebody is hired into, rather than a position somebody arrives at by having started the business., Vaidehi Rao. Not a long roll of appointments. Not a web of entities in six places. A small surround, set out in full.

A short list is information before any amount is attached to it. The ruling comes from where this note is worked in full, and it is the sentence everything here is built on. A list of four lines and a list of thirty entities across four states are two different facts about two different businesses, both free, both readable in twenty seconds, and a reader who skips straight to the figures has thrown away the only part of the note that costs nothing.

Ordinary life does this without thinking. Before asking what anybody ever took, the thing to want is the list of people who can let themselves into the house. Four keys is one situation. Thirty keys is a completely different situation, and that much was clear before anybody mentioned a missing thing.

Two of those lines read none listed. What does that settle, and what does it leave open?

The two empty lines are the centre of the subject, and they repay slow reading. Two of the four lines read none listed, and a line reading none listed is a positive statement rather than a missing entry. The line says that no other entity is controlled by the people who direct this business, and that no close relation of theirs appears on the list at all. Somebody had to look and had to write down what the looking found.

The two empty lines rule out something real and worth having. There is a particular shape money takes when interests diverge, and it is the shape everybody pictures. Money leaves a business toward a company quietly held by the person deciding, or toward a relation of theirs, at a price that ordinary bargaining between strangers would never have produced. The shape is exactly what the note was built to catch, and on this note the two lines where it would show read none listed. One route is closed, on the record, and nothing further needs asking about it.

Now the half a reader almost always skips, and it is this guide's own finding. A divergence of interest does not require a related party, and the note catches one route out of several. Take that apart into cases and it stops being an abstraction.

A manager can prefer a quiet year to a good one. Nothing crosses the boundary of the business, no counterparty exists, and the interest gap is doing real work the whole time. A decision can be taken late, or hesitated over for eleven months, or never taken at all, and no note anywhere in any set of accounts records a decision that was not taken. Money can be spent on things nobody prices separately: more people, more space, a project that sounded good in a meeting. Not one of those needs a connection to exist, and not one of them would ever reach any of the four lines above.

The household version makes the point in one move. The list of people with a key to a house is worth reading, and a short list closes a real question. The key list says nothing at all about whether the person who lives there has been leaving the tap running for a year.

One more sentence has to arrive in the same breath, or a short list reads as an accusation. A four line list with two lines reading none listed is an ordinary and common shape. Most businesses have small surrounds. Anjani Stationers has set out everything the note asks for and has concealed nothing, and there is nothing in the shape of the list that is a criticism of anybody. The finding below is a statement about an instrument, not about a business: the instrument reaches one route, and there are other routes.

Once an amount is attached to a line, which total does it belong against?

The list is read, and now the figures arrive. The division is worked out in full where reading a dealing with a connected party is the subject, so the result and its ruling are quoted here rather than rebuilt.

Chitra Binding Works invoiced Anjani Stationers Rs 8,00,000/- for binding across the year. Anjani Stationers carries more than one figure of exactly that amount in exactly that year, and two of them sit on this one note, so which Rs 8,00,000/- is meant has to be said every single time it is written. The one just named is the binding invoiced, and binding invoiced is a purchase of a service.

Because it is a purchase, it belongs against what the business buys. The total is cost of materials consumedThe total a business spent in the year on the bought-in materials and services that went into what it sold. It is the buying side of the trading account, and it excludes wages, rent and the other costs of simply being open. of Rs 1,48,50,000/-. Rs 8,00,000/- of binding invoiced against Rs 1,48,50,000/- is 5.4 per cent, and that is the total this item belongs to. One division, one sentence, and the step is finished.

Watch what happens if the total is chosen by habit instead. The same Rs 8,00,000/- of binding invoiced, measured against revenue of Rs 2,70,00,000/-, reads 3.0 per cent. Against total expenses of Rs 2,32,00,000/- it reads 3.4 per cent. Against profit before tax of Rs 38,00,000/- it reads 21.1 per cent. One amount, four correct answers, one relevant question. Every one of those divisions is arithmetically right, and the spread from 3.0 to 21.1 is a sevenfold difference produced entirely by choosing a denominator, with no fact anywhere having changed. The same ordinary purchase can be made to look trivial or made to look alarming without a single figure moving.

Two more items sit on the same note and each needs the same treatment. Rs 1,50,000/- of that binding was still owed when the year closed. An amount still owed is a balance rather than a flow, so it belongs against trade payablesWhat a business still owed its suppliers on the last day of the year. It is a position read on one date, not a total that built up across the year. of Rs 22,00,000/-, giving 6.8 per cent. And there is a guaranteeA promise to stand behind somebody else's borrowing if they do not pay. Nothing is paid and nothing is owed while the borrower keeps up, so it is set out in the notes without being recorded as a cost or a liability. of Rs 8,00,000/- standing over Chitra Binding Works' borrowing, disclosed and not recognised. The guarantee is the second Rs 8,00,000/- on this note, and the guarantee is not the binding invoiced. The two are entirely different things that are never added, never netted and never spoken of as one figure.

The guarantee also produces the most useful refusal here. The guarantee belongs against the borrowing it stands behind, that borrowing is Chitra Binding Works' own, and it appears nowhere on the face of these statements. So no proportion is named for it at all. Naming none is the correct output when the total an item belongs to is not on the face in front of the reader, and reaching for a total that happens to be nearby would be exactly the error the previous paragraph is about.

One last line for scale, and then the arithmetic stops. Rs 8,00,000/- of binding invoiced across the 2,50,000 registers the business made in the year is Rs 3.20/- a register. No rate per unit is disclosed for what binding cost before the arrangement changed, so whether Rs 3.20/- is an arm's lengthA dealing settled by ordinary bargaining between two parties with nothing between them. Used here in its plain sense, with no test, no rule and no approval process attached to it. rate is not a question that can be closed here.

What sort of divergence would leave no mark on that note at all?

Take the decisions this business published, one at a time, and ask of each whether the note carries a trace. Five of them are on the record, and the answers are not evenly split.

The binding bought in from Chitra Binding Works at Rs 8,00,000/- across the year: the note carries it, on the line for companies inside the same group, at 5.4 per cent of cost of materials consumed. The Rs 1,50,000/- of that still owed when the year closed: the note carries it, as a balance, at 6.8 per cent of trade payables. The guarantee of Rs 8,00,000/- over Chitra Binding Works' borrowing: the note carries it too, disclosed and not recognised, with no proportion named.

Now the other two. A board of directors approves a machine purchase of Rs 15,00,000/-: the note carries nothing. No connected party is on the other side of it, and the accounts themselves record nothing at all until the machine arrives. Board composition and what a board of directors approves is covered separately in these notes. And the whole Rs 30,00,000/- of the year's profit after tax is retained rather than distributed, with no dividend paid: the note carries nothing there either. Three holders agreed to that, and three holders can change it.

Three of the five leave a mark on that note and two leave none, and the two that leave none are both decisions about where the business's own money goes. Sit with which two those are. Not the small ones. Not the obscure ones. The purchase of a large asset and the use of an entire year's profit are precisely the two the instrument cannot reach.

The shortfall is not a defect in the note and it is not a criticism of anybody who prepared it. The note was built to catch money moving toward somebody connected to the people deciding, and at that job it works. The error is entirely on the reader's side, and it is a simple one: a good instrument pointed at one route has been mistaken for an instrument pointed at everything. The note is a torch, and a torch is genuinely useful. A torch is not daylight, and nothing in the room its beam does not fall on has been ruled out.

So where does that leave somebody reading this business from outside it?

Three things about this business are published, and one is not. Set them down plainly and the position becomes clear, and it is not the position most readers think they are in.

The decisions are published. At the start of its second year Anjani Stationers paid Rs 21,00,000/- for 70 per cent of Chitra Binding Works Private Limited. Chitra Binding Works had been a separate company under separate control. Of the three stages a register passes through, cutting runs 150 an hour, printing 125 and binding 100, so binding governs the pace of the whole works, and binding was the only one bought in. The same business retained the whole Rs 30,00,000/- of its year's profit after tax and paid no dividend.

One outcome is published against them, and it is the flattest sentence in these notes. On the day the Chitra shares moved, binding still ran at 100 an hour and rated capacity was still 4,00,000 registers a year. Holding the stage bought the right to make a change and did not buy the change. The purchase, what it acquired and what the year did afterwards are worked out separately under How Capital Allocation Shapes Long-Term Business Outcomes.

And the people are published, three of them. Anjani Kulkarni decides. Vaidehi Rao is the finance controller who has to justify to her board of directors why profit moved when nothing about the trading did. Meera Rao is the accountant who comes in three days a week. Three names, three roles, all of them on the record.

No line joining a person to a decision, or a person to a holding, is published anywhere. Nobody is recorded as having decided the purchase. Nobody is recorded as having decided the dividend. The decisions are visible, the outcome is visible, and the decider is not named against either.

The honest thing to say about that is not the reassuring one. The missing line is not a failure of these notes and it is not an unusual gap in this particular record. The missing line is the ordinary position of anybody reading a business from outside it, in India and everywhere else, and it is why the whole skill runs on decisions rather than on people. The name against the choice almost never comes. The choice and its consequence very often do, and that is enough to work with once the wait for the other thing is given up.