Asset-Backed Securities: The Structure and the Underlying Pool

An asset-backed security is a claim on a bundle of contracted payments owed by many borrowers, none of them home loans, with the money shared out in a set order. Its machinery copies every other securitisation. Its pool does not. Two things about that pool are worth reading: the count of separate debts inside it, and the weight of its largest one.

Every securitisation runs the same machine around whatever gets fed into it. Once that is settled, the questions worth asking stop being about the machine at all and move to the contents. A count is harder to print than a total, and a largest-item figure is harder still, so the two facts that describe the contents are the two most summaries quietly leave out.

What is actually sitting inside the pool?

Meet Sarvani Receivables Trust, an invented lender's structure. Inside the trust sits Rs 1,200 crore of equipment loan receivables owed by many small businesses, each of them an obligorWhoever carries the duty to pay under one particular contract. In a pool of small business lending, one obligor is one business. under a contract of its own. Somebody bought a bread oven, a lathe, a delivery van; the lender wrote a repayment contract against each; and those contracts, in bulk, are what got moved.

Three pieces fund the structure, and their sizes sit in the table below. Two of them taken together come to Rs 240 crore, and that pair is what the senior piece waits behind. The three pieces behave the same way in every structure of this kind. The pool underneath them is what differs from one structure to the next.

Read the queue column rather than the size column and the arrangement gives itself away. The equity piece stands in front. Behind it waits the mezzanine piece. The senior piece is at the back, and is reached only once both of the pieces ahead of it have been used up. Rs 240 crore, out of a pool measuring Rs 1,200 crore. The ratio of Rs 240 crore to Rs 1,200 crore comes to one rupee in every five, and written out as a percentage it is 20.0.

What has to be true of a receivable before anybody can pool it?

Not every claim on money can be poured into a structure like this one. Four things have to hold, and it is worth walking them because between them they explain why some businesses can raise money this way and others simply cannot, whatever their trade looks like from the outside.

A receivable has to be identifiable, transferable and scheduled, and there have to be a great many of them. Take them one at a time.

Identifiable means the exact set can be written down. Not described, not estimated, not characterised as roughly this much from roughly these customers: listed, contract by contract, with enough detail that a stranger reading the list could say which claims moved and which stayed behind. Everything downstream depends on this. No charge can be registered over a set nobody can enumerate, and no buyer can be told what they have bought if nobody can say what it consists of.

Transferable means the benefit can be handed on without going back to each borrower for permission. The distinction between handing on a benefit and rewriting a contract is sharper than it sounds. If moving a claim required novationTearing up one contract and writing a fresh one with a new party in place of the old, which needs everybody to sign. Slower than simply handing the benefit onward. for every single contract, a pool of four thousand eight hundred claims would need four thousand eight hundred signatures collected one at a time, and the whole idea dies in the collecting. Some contracts are written so this is easy. Some are written so it is impossible.

Scheduled means the money the claim will produce can be described before it arrives. A buyer is being asked to pay today for cash arriving later, and unless the shape of that cash can be set out in advance there is nothing to price and nothing to divide into pieces.

Many of them is the test people forget, and everything else here grows out of it. One receivable moved into a structure of its own is not a securitisation; it is a loan wearing a great deal of paperwork. The pooling is not decoration on top of the transfer. The pooling is where the work happens.

Put the four together and they explain a pattern often noticed without a reason attached to it. A business that lends in standard written form, in bulk, on stated dates, can raise money this way almost as a matter of routine. A business owed a great deal of money on running accounts, with no written repayment dates and terms settled by conversation, cannot, however sound its customers are and however reliably they pay. The obstacle is not the quality of what it is owed. The obstacle is the form the debt was written in, and the form was chosen years before anybody thought about raising money against it.

The inversion runs against the order most people expect, and it is worth sitting with. The natural assumption is that the question is whether the borrowers are good. The four tests never ask. Their question is whether the claims can be listed, moved, described in advance, and counted in large numbers. A pool of weak borrowers passes all four. A pool of excellent ones written up carelessly fails the first two. Quality is a separate question, asked separately, and it is not what decides whether a structure can exist at all.

Which facts about a pool actually change what is owned?

A description of a pool arrives. Some of what it says moves the answer to every question that matters, and some of it is furniture. Sorting the two is a skill, and there is a single test that does the sorting.

The total moves the answer, obviously. So does the schedule. The schedule decides when the money turns up rather than merely how much of it there is. And then two more that summaries usually skip: how many separate receivables the pool holds, and how large the largest single one is. Everything that follows in this guide works on those two.

On the other side sit the facts that read like information and do nothing. The originator's name is one. A general description of the trade the borrowers are in is another. So is anything printed about the pool that is neither a number nor a date. None of these is dishonest; none of them is useless in every context. None of them can enter a calculation, and a fact about a pool earns attention exactly when it can be put into a sum.

Why does the number of receivables matter as much as the total?

Take that same Rs 1,200 crore and hold it two ways. Both compositions are declared rather than counted, and a summary printing only the total cannot tell the two apart.

Composition one: 4,800 receivables averaging Rs 25,00,000/- each. Multiply out and the pool comes to Rs 1,200 crore. Now let one of them fail completely and return nothing. Rs 25,00,000/- comes out of an equity piece of Rs 60 crore. The failure disappears inside it. The equity piece absorbs the whole thing without difficulty, the mezzanine piece is not touched, and the senior piece never hears about it.

Composition two: 8 receivables of Rs 150 crore each. Multiply out and the pool comes to Rs 1,200 crore, exactly as before. Let one of these fail completely. The equity piece is Rs 60 crore, so Rs 60 crore is the most it can take. Strip that off a failure of Rs 150 crore and Rs 90 crore is left over with nowhere yet to go, so the mezzanine piece takes it. Divide Rs 90 crore into a mezzanine piece holding Rs 180 crore and exactly half of that piece has gone: 50.0 per cent.

Stand those two answers next to each other. Same pool total. Same three pieces. Same queue. One borrower failing, and the second reading is that half the mezzanine piece has gone while the first reading is that essentially nothing happened. Read the queue first, then read what is standing in it. The queue alone does not say how far one name can travel up it.

The arithmetic is the whole finding here, and it is worth doing rather than reading. Walk the second case through the queue slowly.

Rs 150 crore fails and returns nothing. The equity piece stands in front, and it holds Rs 60 crore, so Rs 60 crore of the failure stops there. Take the Rs 60 crore away from the failing Rs 150 crore and Rs 90 crore is what remains. The remaining Rs 90 crore has to go somewhere, and the next piece in the queue is the mezzanine piece at Rs 180 crore. The mezzanine piece takes the Rs 90 crore, and Rs 90 crore is half of Rs 180 crore. Rs 240 crore stood in front of the senior piece and Rs 150 crore of that Rs 240 crore has now been consumed, so Rs 90 crore is still standing there and the senior piece absorbs Rs 0/-.

Take the control all the way across and one cell sits there unchanged. The senior piece absorbs Rs 0/- at every setting, all ten of them. Rs 240 crore stands in front of it and the control tops out at Rs 150 crore. A readout that refuses to move across an entire range says something a moving one cannot: the distance is not close. Why the range tops out there is worth saying plainly. Rs 150 crore is the largest single receivable inside the declared eight-receivable pool, and pushing beyond it would mean drawing a pool nobody has declared.

Is the word backed saying anything about security?

It is saying something, and the something is narrower than most readers assume. Backed says that the pieces are claims on a defined set of receivables rather than on a business in general. The narrowing is real, and it matters. Whatever else the originator does after the transfer, well or badly, a holder's outcome hangs on that particular set of contracts and not on the rest of its affairs. The set was identified, it moved, and it is now the thing the claim runs against.

The word leaves a longer list undone. Backed does not say the receivables are secured on anything, it does not say they will be paid, and it does not rank this instrument against any other. Whether a given pool happens to be secured on the equipment it financed, or on nothing at all, or on something in between, is a separate fact about that particular pool, written into the documents, and it varies from one structure to the next. Two instruments can both be honestly called asset-backed and differ completely on that point.

The confusion is understandable. In ordinary English the word asset does a lot of quiet work. In this name the asset is the receivable itself. A contracted payment is an asset to whoever is owed it. The claim ends there, and reading it as a promise about what stands behind the payment is a step the name never took.

Bring it back to this pool and the point sharpens. Sarvani Receivables Trust holds payments owed against equipment that was bought with the money. Somebody could have written those contracts so that the oven and the lathe stand behind the payments, and somebody could have written them so that they do not. Both versions would be described, correctly, as asset-backed. The name of the instrument cannot tell the two apart, so which of the two a holder has is settled by the documents and not by the name. Which way Sarvani was written is a fact about the documents, and no document exists.

What stays exactly the same whatever the pool holds?

Almost all of it, and this is the part worth learning once rather than relearning per structure. The pool sits in a structure of its own. One party moved it across. A second brings the money in as it lands. A third stands over the structure for whoever bought pieces of it. The three pieces are a queue rather than shares of the pool. A loss lands on whichever piece stands in front and works its way back. The senior piece has Rs 240 crore ahead of it, a fifth of everything in the pool.

Not one of those sentences has to be rewritten because the receivables are equipment loans instead of something else. The machine is worth learning properly the first time for exactly that reason. Whether the pool holds home loans or equipment loan receivables, the structure still holds Rs 1,200 crore funded by three pieces standing in an order, with Rs 240 crore in front of the senior piece. Where the two kinds of pool part company is a subject of its own, and it is not opened here.

Where does this shape already appear, without being called granularity?

Two shopkeepers shut their books at the end of the month. Each is owed Rs 80,000/-, and each writes that figure at the foot of the ledger.

The first is owed it by three hundred and twenty customers who each took away a Rs 250/- item on credit. The second is owed it by two customers who each ordered Rs 40,000/- of goods. Put the two ledgerThe running record of who owes whom, kept line by line. A shopkeeper notebook and a bank core system are the same object at very different scales. sheets side by side and the totals match to the rupee. There is nothing to choose between them if the total is all anybody looks at.

Now let one customer refuse to pay. The first shopkeeper is out Rs 250/-, a rounding error on a month. The second is out Rs 40,000/-, half the month gone. Rs 40,000/- against Rs 250/- is a hundred and sixty times the damage, and no summary of either sheet that printed only the total would let anybody tell the two apart. The count is the missing row, and it is missing in both directions: it would have shown that the first shopkeeper is safe and the second is exposed.

The shape shows up elsewhere once it has been recognised once. A household running on one salary against a household running on three. A food stall outside a single office building against one on a busy junction. Ten shops in one mall sharing one stream of visitors. The total takings can be identical in every pair, and what one bad month does to each is not.

How does anybody actually use this in a working week?

A lender that originates receivables and moves them on reads the count from the other side of the table. The lender knows what it wrote, contract by contract, so the composition is not news to it. A pool that behaves well under a single failure is only worth assembling if somebody can see that it does, so the lender watches whether the pool it assembled will be read correctly by the people buying pieces of it.

An analyst handed a pool description does the opposite job, and the first move is a subtraction rather than a calculation: what is not on this sheet? If the count is absent, the largest-receivable figure almost certainly is too, and everything below then rests on a total that two very different pools could produce. The request that follows is short and always the same shape. How many, and how big is the biggest.

Somebody holding a piece and marking it period after period cares about the count for a reason the analyst does not: it decides how much any single piece of news matters. In a pool of thousands, one business closing is weather. In a pool of eight, one business closing is the whole story, and there is no averaging to hide behind.

The servicer, collecting month after month, meets the same fact as workload rather than as arithmetic. Chasing 4,800 payments is an operation: systems, reminders, a process that runs whether or not anybody is thinking about it. Chasing eight is eight conversations, and the person having them knows every name. Neither is harder in some absolute sense. The difference is that in the second case one conversation going badly is a material event for everybody upstream, and in the first case no single conversation is.

And a household reads this without ever meeting the word. One earner in a house, or three earners in the house next door, and the same amount can land in each account every month. The two are separated not by the total that arrives but by how many separate sources it arrives from. Concentration in a household is the same question in a smaller currency. Unease about a single client paying most of the bills is already an understanding of concentration; this guide only supplies the arithmetic for it.

What does a full description of the pool still not say?

Everything about what sits inside the pool has now been set out. How the pool then behaves has not, and the gap between those two things is wider than it looks from here.

Nobody has supplied a rate at which these receivables stop paying. Nobody has supplied what comes back when one does. And nobody has said anything at all about whether they would fail together or independently. Failing together is a third question again, and the first two answers do not settle it. Not one of the three rates exists anywhere in the material. The arithmetic that survives all three absences is the arithmetic of a loss once a loss has happened: how far up the queue it reaches. Which loss to expect is a different matter. The arithmetic for that is absent, and being very careful with the arithmetic that is present will not conjure it.

The distinction is easy to slide off, so hold onto it. One receivable of Rs 150 crore has just taken half the mezzanine piece. A short step from there is a feeling about which composition is preferable, and that step needs a fact nobody has supplied. Preferring one would need to know how often a large receivable falls into arrearsMoney that was due and has not turned up yet. A payment can be in arrears for a week or for a year, and the word says nothing about which. and stops for good compared with a small one, and nothing on this platform speaks to that comparison.

So the queue is a statement about who absorbs before whom, and nothing else. The queue says who is in front. The queue says nothing about how often anybody in front actually gets hit. No likelihood is attached to any of the three, and none of them is assessed. The word safe belongs to none of them either. Safety would be a claim about how often, and how often is a fact the material does not hold.

Sarvani Receivables Trust and the small businesses behind its receivables are invented.
Educational material. Not advice on any investment, tax, budget or market position.