Securitisation: Turning Loans Into Tradeable Securities

Securitisation lifts a pool of loans off one lender's books and into a structure built for that pool alone. The new structure funds the pool by issuing pieces paid in a settled sequence. The pieces are not fractions of the pool. A loss strikes the lowest piece, then climbs, so what a holder actually has is a place in a line rather than a slice of anything.

A lender sitting on a few thousand loans is sitting on one lumpy thing that nobody else can price and nobody else can buy. Move those loans into a structure that exists for nothing but that pool, and the very same cash flows can be funded by instruments differing from each other in exactly one respect: which of them meets a loss before the others do. The ordering of the pieces is what makes them different objects, and that ordering is the whole of the subject. Everything below is the same four amounts read in different directions.

What are the four amounts, and does the addition close?

Four numbers can be held in mind before an idea can, so the arithmetic comes before the abstraction. Sarvani Receivables Trust, an invented structure, holds a pool of receivablesAn amount somebody has contracted to pay on a stated date. A pool here is made of thousands of them, and what each borrower promised was settled long beforehand. worth Rs 1,200 crore. Three pieces fund that pool, and nothing else does. Here they are, with what each is worth reading twice over.

The addition is worth doing independently rather than trusting the last row. Rs 960 crore, plus Rs 180 crore, plus Rs 60 crore on top of that, comes to Rs 1,200 crore. The pool is funded to the rupee, which also means the structure keeps no cushion of its own: every rupee that funds the pool belongs to one of the three pieces, and the structure has nothing else to fall back on. The right-hand column is the same addition converted, so it closes the same way, at 100.0 per cent.

The addition is worth holding in mind. The addition is correct. It closes both ways, in rupees and in per cent. And on its own it leads to a conclusion that is flatly wrong.

Notice what the drawing leaves untouched. No borrower changed. No instalment moved. No rate was rewritten. The receivables are the same receivables, promising the same amounts on the same dates, and a borrower paying into this pool need never learn that any of it happened. The change is in who is entitled to the money as it arrives, and in what sequence.

Why do the receivables have to leave the lender at all?

Suppose the lender kept the loans and simply borrowed against them. A buyer handing over money would then be exposed to two entirely different things at once: the borrowers in the pool, and the lender itself. If the lender ran into trouble over something wholly unconnected with these loans, the buyer would discover that they had bought a claim on a business rather than a claim on a set of receivables.

So the receivables are moved into a structure that exists for this pool and for nothing else, holds no other business, and takes on no other obligation. The single-pool structure is not decoration. It is the point. A structure with one pool inside it and no other activity holds nothing else to assess, so a buyer can actually assess it. Every question a buyer has narrows to a question about the receivables and about the order in which the pieces are paid.

The everyday version runs like this. A friend who asks for a loan against the rent from one shop they let out is asking the lender to lend against the rent and against the friend, both at once. If instead the shop and its lease are put into a separate arrangement that holds nothing but that shop, the lending is against the rent alone. The second is easier to price, not because the rent got safer, but because there is less to think about.

Whether a particular transfer actually achieves that separation is a question with a formal answer, and that answer is settled elsewhere. Whether the receivables have genuinely left the originatorWhoever did the lending, and then sold what was owed into the structure. How that role works is covered separately.'s own books is decided by a test the Institute of Chartered Accountants of India publishes at icai.org. Which receivables may be moved at all is set by the Reserve Bank of India at rbi.org.in. Registration of the chargeA claim over a named asset, put on a public record. Its whole purpose is priority: whoever registered earlier stands ahead. Where and how it is recorded is the registry's business, not either party's. over the receivables is lodged at cersai.org.in with the central registry. All three get revised, and the wording stands where it is published.

What does reading the pieces as shares actually get right?

Take the three amounts and read them as shares of the pool. Almost everybody does that on first meeting. Rs 960 crore of Rs 1,200 crore is 80.0 per cent. Rs 180 crore is 15.0 per cent. Rs 60 crore is 5.0 per cent. The three shares look like a pie cut into three slices, one big, one middling and one thin.

The shares reading is correct about exactly one thing: how much money each holder put in. It is not a lie and it is not a rounding error. Asked which holder wrote the biggest cheque, the shares reading answers perfectly, and no other reading answers it better. The trouble starts the moment a second question is put to the pie.

A pie has no order in it. Ask it who meets the first rupee of a loss and the pie has nothing to say. Slices of a pie are simultaneous. The slices all exist at once and none of them comes before another. The three pieces of Sarvani Receivables Trust are not simultaneous at all, and the difference is not a detail.

Where does the shares reading break?

The three pieces do not share a loss in proportion. The pieces meet a loss one after another. A loss on the pool lands on the equity piece first, and it reaches the mezzanine piece only once the equity piece has been used up, and it reaches the senior piece only once the mezzanine piece has been used up as well. The order of that landing is the mechanism. Everything else is arithmetic run on it.

Read by size, the equity piece accounts for 5.0 per cent of the money, and read by order it takes 100.0 per cent of the first Rs 60 crore of loss. Sit with how odd that pairing looks written down. The smallest holder, by a wide margin, is the only holder exposed to the beginning of any trouble. The addition was never about loss in the first place. Nothing in the shares reading hints at the pairing, and nothing in the addition contradicts it.

Now the discipline that keeps this straight, and it is the single most useful habit in the whole subject: name the base of every percentage, every time. Rs 60 crore of loss reads as 5.0 per cent when the pool is the base. That very same Rs 60 crore reads as 100.0 per cent when the equity piece is the base. The two readings are not two events. They are one event described against two different bases, and an account that drops the base leaves two numbers with no way to tell them apart. Whenever a percentage appears in this material, the first question is what it is a percentage of.

Two arrows, running against each other, are the whole structure. Money collected from the borrowers enters at the top and falls: the senior piece is satisfied, then the mezzanine piece, then whatever remains reaches the equity piece. Loss enters at the bottom and climbs: the equity piece is consumed, then the mezzanine piece, and only after both are gone does anything reach the senior piece. A holder of the equity piece is therefore last in line for money and first in line for trouble, and both of those follow from the one ordering.

What does the order buy, in one number?

Everything above has been qualitative: first, then, only after. Turned into an amount, it becomes something that can be checked. Under the senior piece sit the other two, one on top of the other. Rs 60 crore of equity piece. Rs 180 crore of mezzanine piece. Rs 240 crore between them, and the pool reads that as 20.0 per cent.

Rs 240 crore is the amount the pool can lose before the senior piece loses a single rupee. That figure is not a promise, not a probability and not a quality of the senior piece. Rs 240 crore is an amount, sitting there, that anyone can add up from two of the four numbers. It is also, and this is worth being blunt about, the only thing the order gives anybody.

Run the same reading down the stack. Under the mezzanine piece there is only the equity piece, Rs 60 crore of it, and the pool reads that as 5.0 per cent. Under the equity piece there is nothing whatever. No cushion, no anything, Rs 0 crore. The equity piece is where the structure starts.

There is a second way to look at the same three amounts, and it makes the structure fall out as a shape rather than a list. Put pool loss on a scale from nought to a hundred per cent and ask, at each point along it, which piece is absorbing. The equity piece absorbs from 0.0 to 5.0 per cent. The mezzanine piece absorbs from 5.0 to 20.0 per cent. The senior piece absorbs from 20.0 per cent onwards. The bands meet exactly, with no overlap and no gap between them, because each band is exactly as wide as its piece is big.

The brackets above the bar say the same thing in width: 5.0 points of the scale for the equity piece, 15.0 for the mezzanine piece and 80.0 for the senior piece, adding to the whole hundred. Which is the shares reading again. The shares reading was never wrong about size; it was only ever silent about position, and the band picture carries both at once.

What happens to each piece as the pool loses more?

Four positions are enough to show the whole shape, and every one of them is a subtraction that can be done on paper. Each starts from a percentage of the pool, converts to rupees, and then walks those rupees down the order one piece at a time.

Start at 5.0 per cent. On Rs 1,200 crore that is Rs 60 crore. All of it lands on the equity piece. The mezzanine piece takes nothing. The senior piece takes nothing. The wipe is exact, and it is exact for the plainest possible reason: Rs 60 crore is what the equity piece is. Not approximately. Not roughly. The same number appears on both sides of the sentence because the record put it there.

Move to 12.0 per cent. Rs 144 crore. The equity piece is already gone at Rs 60 crore, so Rs 84 crore is left over, and the mezzanine piece takes all of that. Set Rs 84 crore beside a mezzanine piece sized at Rs 180 crore and the reading is 46.6667 per cent. The senior piece is still untouched. Two different bases have now appeared in two sentences: 12.0 per cent was of the pool, and 46.6667 per cent is of the mezzanine piece.

Move to 20.0 per cent. Rs 240 crore. The equity piece takes Rs 60 crore, the mezzanine piece takes Rs 180 crore, those two come to exactly Rs 240 crore, and the senior piece is still untouched. A pool loss of 20.0 per cent is the last position at which that sentence is true, and the amount is the same Rs 240 crore that stands beneath the senior piece. It is the same fact approached from the other end.

Move to 28.0 per cent. Rs 336 crore. Subtract the Rs 240 crore standing beneath and Rs 96 crore is left for the senior piece. Set that Rs 96 crore against the senior piece's own Rs 960 crore and the reading is 10.0000 per cent. So a pool that has lost more than a quarter of itself has cost the senior holder a tenth of their money, and reading those two figures as though they were comparable is exactly the mistake the base rule exists to prevent.

Who runs the structure once it exists?

Three jobs, usually three different parties, named here and taken no further. Somebody made the loans and sold them into the structure. Collecting the money every month and passing it on is the servicerThe party that keeps collecting instalments from the borrowers after the receivables have moved, and hands what it collects to the structure. Often, though not always, the same party that made the loans.'s job. Standing in for whoever bought the pieces, holding the structure on their account and acting when things stop running as written is the trusteeA party that holds something for others rather than for itself. Here it is the one with standing to act if the arrangement misbehaves, and the buyers are who it acts for.'s job.

Notice that none of those three jobs is the job of deciding who absorbs a loss first. The order was settled before any of them started work. The order is written into the documents when the structure is set up. The parties running it execute the order; they do not choose it, and they cannot change it because a particular month has been difficult.

The duties of those three parties are a different matter, settled elsewhere and rewritten periodically. The Reserve Bank of India at rbi.org.in and the Securities and Exchange Board of India (SEBI) at sebi.gov.in are where those duties live. Who does what, in detail, is covered separately.

What does this look like with the rupees taken out?

Ten shops in one mall put a month of card takings into a single box. Three people are promised money out of that box in a written order. The first is paid in full before the second gets anything. The second is paid before the third gets anything. The third is paid only out of whatever is left over once the first two are done.

Now suppose the takings come up short. The shortfall does not get divided by three. The shortfall falls on the third person until there is nothing left of their claim, and only then does it touch the second. Nobody in that arrangement holds a third of the box; each of them holds a place in a line, and the place is the whole of what they hold.

How does somebody actually use any of this?

Start with a lender. A lender that has made a few thousand small loans has a fixed amount of capacity, and every rupee still sitting on its books is a rupee it cannot lend to somebody else. Moving a pool into a structure frees that capacity, so the reason a lender does this is usually plain arithmetic about how much it can lend next quarter rather than anything clever. Whether the capacity is actually freed, in the sense a regulator recognises, depends on the accounting test and on the capital treatmentThe amount of a lender's own money a supervisor requires it to hold against a given exposure. What that amount is, and when a transfer changes it, is set by the supervisor rather than by the two parties to the transfer., and both are routed below rather than stated.

Now an analyst looking at a piece of a structure like this. The first thing worth working out is not a rate and not a return. The amount standing beneath a given piece comes first. The size of that amount sets what that piece is exposed to, and everything else is read against it. On this structure that is Rs 240 crore for the senior piece, Rs 60 crore for the mezzanine piece and nothing for the equity piece. An analyst who cannot state that amount for a piece has not started work on it, whatever else they have calculated.

Then a household. The shape turns up outside finance too. A joint arrangement where one earner covers the rent in full and the others cover whatever is left is a queue, not a split. If money is short one month, it does not come off everybody equally. The shortfall comes off whoever was last in the written order. The habit worth taking away is small and portable: whenever somebody describes how something is divided, the separate question is what happens when there is less of it than expected. The two answers are often unrelated, and only the second one is about risk.

Finally, what none of those three people can get from the four amounts. A lender's decision to do the transaction, an analyst's view on a piece and a household's arrangement all turn at some point on how likely a shortfall is. How likely a shortfall is cannot be answered from these four amounts, and that question is taken up below.

What goes wrong when the pieces are read as a pie?

The gap on the equity piece alone is Rs 57 crore, twenty times the figure the proportional reading gives it. A gap that size is not a rounding disagreement or a modelling choice. It is two different questions being answered by two different people who both think they are answering the same one.

What can this structure not settle?

Most readers came for the likelihood, and the likelihood is where the arithmetic stops. Nothing above says how often a pool of this kind loses money, or how much. Not once. The order in which a loss lands is set out completely, and the account ends there.

The four amounts settle who absorbs before whom, exactly and completely. The four amounts settle nothing whatever about whether anything is absorbed at all. Putting a chance on any of the three pieces would take two things that do not exist in this material: a spread of possible pool losses, and a view on whether the receivables in the pool fail together or one at a time. Holding the certainty and the ignorance side by side is the actual skill.

Assessment is settled the same way. None of the three pieces of Sarvani Receivables Trust carries a grade, and a grade is not something four amounts can produce. Scales belong to the firms that build them. Each firm writes its own steps. Each explains those steps in its own document. The disclosure required about such an assessment sits with SEBI at sebi.gov.in. A row left blank is a more honest object than a row filled with a grade nobody published.