How loan pools become securities: SPVs, pass-throughs, prepayment, tranches, waterfalls, credit enhancement, CLOs and default correlation, worked step by step.
By Prof. V. Ravichandran · Fixed Income Securities · CFA / FRM / MBA · 24 learning sections · two Apex case studies · 13-sheet Excel companion · Interactive waterfall simulator
Pool · SPV · tranches · waterfall · losses from the bottom up
₹100 CR
Cash flows down the structure; losses climb up it.
Senior notes are paid first and lose last; equity is paid last and loses first.
Pass-through month
₹0.6071 cr
₹0.50 cr interest + ₹0.107 cr principal
12% CPR as SMM
1.0596%
SMM = 1 − (1 − CPR)1/12
Weighted average life
2.3 years
Last payment in year 3
Senior protection
20%
₹80 cr senior over ₹20 cr junior
CDO equity cash yield
23.0%
Apex CDO I, no-default year; not a promise
Senior loss, correlated
₹8 cr
vs ≈ ₹0 if defaults independent
FOLLOW THE CASH
Borrowers → servicer → trust accounts → waterfall → investors, after fees and in strict priority.
Investor money buys the pool at closing; borrower repayments pay investors afterwards.
THE TEACHING POINT
Tranching redistributes risk; it does not remove the borrowers’ credit risk.
Use the simulator to watch losses travel up from equity to the senior class.
Hypothetical teaching data · amounts in ₹ crore · every example recomputed
01 / PURPOSE
What is securitisation, and why do lenders use it?
Securitisation pools financial assets and issues securities whose payments depend on the cash those assets generate. A bank that holds ₹100 crore of housing loans can transfer them to a trust; the trust issues securities to investors; homeowners’ repayments then pay those investors. The lender turns an illiquid loan book into cash, and investors buy exposure to a defined pool rather than to the whole bank.
Typical collateral includes home loans, vehicle loans, credit card receivables, equipment loans and leases, and corporate loans. Lenders securitise to raise funding, improve liquidity, manage maturity mismatches, diversify funding sources, transfer some credit risk and free capacity for new lending.
Four questions for every securitised product
Who pays?The underlying borrowers.
What cash is collected?Interest, principal, prepayments, recoveries.
CPR or ABS speed (US auto-ABS convention: prepayments as a % of the original number of loans each month), cumulative net loss
Default rate (CDR), recovery, OC and IC tests
Terminology. “ABS” is also an umbrella term that can include MBS and CDOs. In market conversation it usually means securities backed by non-mortgage receivables, which is how it is used here.
Remember the starting point: MBS → mortgages · ABS → receivables such as vehicle or consumer loans · CDO → a debt portfolio, then ask how its credit risk is divided.
03 / PARTICIPANTS
Who does what in a securitisation?
Participant
Main responsibility
Borrower
Repays the underlying loan
Originator
Makes the loans and sells the pool
SPV or trust
Holds the transferred assets and issues the securities
Arranger / investment bank
Designs the structure and places the securities
Servicer
Collects repayments, manages arrears and recoveries
Trustee and paying agent
Oversee the documents and apply the payment waterfall
Rating agency
Gives an opinion on the credit risk of rated classes
Investor
Funds the transaction by buying the securities
Guarantor, where used
Provides specified payment protection
04 / MONEY FLOWS
How the money moves: two separate flows
The underwriter is not in the payment chain once the deal has closed.
At closing, investors pay for the securities, the SPV receives the proceeds and uses them to buy the pool from the originator. Over the life of the deal, borrowers pay the servicer, collections go into designated transaction accounts, fees and other required amounts are deducted, and investors are paid according to the contractual waterfall.
05 / SPV
The SPV, true sale and bankruptcy remoteness
An SPV separates the securitised assets from the originator’s other business. Investors buying securities backed by housing loans want exposure to those loans, not to every risk the originating bank runs. A properly structured true sale seeks to move the assets out of the originator’s insolvency estate, so that the originator’s creditors cannot reach them if it fails.
06 / MBS
Mortgage-backed securities: RMBS, CMBS, agency and private label
An MBS is a security backed by a pool of mortgage loans. A lender pooling 1,000 housing loans with an average balance of ₹10 lakh creates a ₹100 crore pool (1,000 × ₹10 lakh). Investors hold securities issued under the transaction; they do not each become the registered owner of a borrower’s house.
Feature
RMBS
CMBS
Property financed
Houses and residential apartments
Offices, hotels, warehouses, shopping centres
Repayment support
Household income
Rent and other property income
Key credit factors
Borrower income, loan-to-value, employment
Occupancy, rental income, debt-service coverage
Concentration
Many smaller loans
Often fewer, larger loans
Are all MBS government-guaranteed? No. In the United States, MBS may carry agency-related guarantees or be issued as private-label securities with no guarantee. Ginnie Mae’s guarantee carries the explicit full faith and credit of the US government; Fannie Mae and Freddie Mac are government-sponsored enterprises with their own guarantee arrangements, which should not be described as identical to Ginnie Mae’s. Even a fully guaranteed MBS still exposes investors to interest-rate risk, prepayment risk and price changes.
07 / PASS-THROUGH
Pass-through cash flows: a worked month
A pass-through security distributes the pool’s collections, after contractual deductions, to investors in proportion to their holding: interest, scheduled principal and unscheduled principal (prepayments). Principal repayment is a return of capital, not interest income.
An investor holding 1% of the certificates receives ₹50,000 of interest and ₹10,707 of principal: ₹60,707 in total. Prepayment is tiny in month 1 because new loans prepay slowly; the next section shows it growing as the loans season. Real transactions add payment delays, reserves, defaults and day-count conventions.
08 / PREPAYMENT
Prepayment risk: CPR, SMM, PSA, contraction and extension
Prepayment risk is the risk that borrowers repay principal earlier than expected, because they refinance at a lower rate, sell the property, use savings to cut debt or move the loan to another lender. When rates fall and refinancing accelerates, principal returns early and must be reinvested at lower yields.
Risk
Typical rate move
Borrower behaviour
Investor consequence
Contraction risk
Rates fall
Refinancing accelerates
Principal returns earlier, reinvested at lower yields
Extension risk
Rates rise
Refinancing slows
Principal stays out longer at an old, lower coupon
Prepayment speed is quoted as a CPR (conditional prepayment rate, annualised) and converted to a monthly rate, the SMM (single monthly mortality):
The PSA benchmark captures the fact that new loans prepay slowly at first. At 100% PSA the CPR is 0.2% in month 1 and rises by 0.2% a month to 6% at month 30, then stays at 6%; 200% PSA doubles every rate. For a ₹100 crore pool at a 6.5% loan rate and a 6.0% pass-through rate, 100% PSA gives:
Month
Opening balance
CPR
Scheduled principal
Prepayment
Net interest
Total cash
1
100.0000
0.2%
0.0904
0.0167
0.5000
0.6071
2
99.8929
0.4%
0.0909
0.0333
0.4995
0.6237
6
99.2936
1.2%
0.0926
0.0998
0.4965
0.6889
12
97.8871
2.4%
0.0949
0.1978
0.4894
0.7821
Year 1
1.1132
1.2924
5.9459
8.3515
Amounts in ₹ crore; closing balance after 12 months ₹97.5944 crore. Scheduled principal is re-amortised on the current balance each month. CPR measures prepayment, not default: defaults are tracked separately (CDR).
Pool averages and loan age: gross WAC, WAM and WALA
A pool’s rate and term are balance-weighted averages, because a ₹30 crore loan group matters more than a ₹10 crore one. The 6.5% pool above could be built from five loan groups:
Loan age (seasoning) is the number of months since a loan was originated: remaining term = original term − age. The PSA ramp runs on loan age, not on the month of the deal. The table above assumes brand-new loans, so deal month 1 is age 1. A pool already 24 months old enters the deal at age 25, so at 100% PSA its first-month CPR is 6% × 25 ÷ 30 = 5.0%, not 0.2%, and its WAL is shorter.
Defaults use the same arithmetic. The conditional default rate (CDR) is an annualised default rate, converted to a monthly default rate (MDR) exactly as CPR converts to SMM: MDR = 1 − (1 − CDR)1/12. A 3% CDR is about 0.2535% of the balance defaulting each month.
09 / CONVEXITY
Negative convexity and the option-adjusted spread
Illustrative shapes, not a pricing model.
For many conventional fixed-rate pass-throughs, falling rates speed up refinancing and cap price gains, while rising rates slow prepayments, extend duration and deepen losses. The price therefore rises less when yields fall than a comparable option-free bond would: negative convexity. Economically, homeowners own a valuable option to repay early and investors are short that option. Not every MBS tranche has the same convexity.
The option-adjusted spread (OAS) is the spread, in a model that simulates rate paths and prepayments, that makes model value equal to price after removing the cost of the prepayment option. It helps separate compensation for credit, liquidity and model risk from the option, but its value depends on the rate and prepayment models used. A higher OAS is not automatically a better investment if assumptions or other risks differ. For the effective duration and convexity used in practice, see Duration & Convexity, Section 18.
10 / AVERAGE LIFE
Weighted average life and sequential-pay CMOs
Weighted average life (WAL) measures the average time until principal is returned, weighted by the amount repaid.
WAL = Σ(t × principal repaid at t) ÷ total principal
Year
Principal repaid
Year × principal
1
₹20 crore
20
2
₹30 crore
60
3
₹50 crore
150
Total
₹100 crore
230
WAL = 230 ÷ 100 = 2.3 years, although the last payment is in year 3.
A collateralised mortgage obligation (CMO) divides mortgage cash flows into classes with different payment rules. In a simple sequential-pay CMO, Class A (₹60 crore) receives all principal until it is retired; only then does Class B (₹40 crore) receive principal.
Year
Pool principal
To A
To B
Closing A
Closing B
1
₹20 crore
₹20 crore
Nil
₹40 crore
₹40 crore
2
₹25 crore
₹25 crore
Nil
₹15 crore
₹40 crore
3
₹55 crore
₹15 crore
₹40 crore
Nil
Nil
Class A’s WAL is (20 × 1 + 25 × 2 + 15 × 3) ÷ 60 = 1.92 years; Class B’s is 3.0 years. Class A absorbs contraction risk; Class B carries more extension risk.
11 / ABS
Asset-backed securities: auto loans, cards, student loans and leases
In common usage, ABS are securities backed by pools of non-mortgage receivables. Eligibility depends on local rules: the examples below are international.
ABS type
Collateral
Cash-flow profile
Key performance measures
Auto loan ABS
Retail vehicle loans
Amortising monthly instalments
Default rate, recoveries, cumulative net loss, prepayment speed
Auto loan example. A finance company pools 10,000 vehicle loans averaging ₹1 lakh: ₹100 crore of principal. The pool moves to an SPV, which issues securities; owners keep paying instalments to the servicer. The main risks are defaults, weak vehicle recoveries, prepayments, fraud and servicing disruption. The resale value of repossessed vehicles matters because it decides how much is recovered after a default.
Cumulative net loss and the performance trigger
Auto ABS investors watch cumulative net loss (CNL): defaults minus recoveries, added up since closing, divided by the original pool balance (not the current balance). Deals set a rising trigger schedule; if CNL crosses it, cash that would have gone to junior classes is redirected to the senior notes. For the ₹100 crore auto pool, with a 45% recovery on repossessed vehicles (counted in the same quarter for simplicity):
Quarter
Gross defaults
Net loss (55%)
CNL, base
Trigger
Base
CNL if defaults double
Stressed
1
0.40
0.220
0.22%
1.0%
Pass
0.44%
Pass
2
0.70
0.385
0.61%
1.0%
Pass
1.21%
Breach
3
0.90
0.495
1.10%
2.0%
Pass
2.20%
Breach
4
1.00
0.550
1.65%
2.0%
Pass
3.30%
Breach
5
0.90
0.495
2.15%
3.0%
Pass
4.29%
Breach
6
0.80
0.440
2.59%
3.0%
Pass
5.17%
Breach
7
0.70
0.385
2.97%
4.0%
Pass
5.94%
Breach
8
0.60
0.330
3.30%
4.0%
Pass
6.60%
Breach
Net loss = gross defaults × (1 − recovery) · Quarter 4: 1.00 × (1 − 0.45) = ₹0.55 crore
CNL = Σ(defaults − recoveries) ÷ original pool = (0.220 + 0.385 + 0.495 + 0.550) ÷ 100 = 1.65%, below the 2.0% trigger
Amounts in ₹ crore. In the base case CNL reaches 3.30% after two years and never touches the trigger. If defaults double, CNL is 1.21% by quarter 2, the trigger trips, and the deal switches to protecting the senior class long before any senior principal is at risk. Real deals add recovery lags and delinquency triggers.
Revolving structures. Card ABS do not amortise at first. During the revolving period investors receive interest while principal collections buy new receivables. If performance deteriorates, an early-amortisation trigger stops reinvestment and starts repaying investors. A common trigger is the three-month average excess spread falling below zero:
Month
Portfolio yield
Charge-offs
Excess spread*
3-month average
Status
3
17.8%
6.2%
4.1%
5.40%
Revolving
5
16.9%
8.8%
0.6%
2.37%
Revolving
6
16.5%
9.5%
−0.5%
0.83%
Revolving
7
16.0%
10.5%
−2.0%
−0.63%
Early amortisation
*Excess spread = portfolio yield − investor coupon (5.5%) − servicing fee (2.0%) − charge-offs. One negative month does not trip the trigger; a negative three-month average does.
12 / TRANCHES
Tranches and the cash-flow waterfall
A tranche is a class of securities with specified payment and loss-allocation rights. Consider a ₹100 crore pool financed as:
Tranche
Amount
Position
Senior
₹80 crore
Paid first; absorbs losses last
Mezzanine
₹15 crore
Intermediate
Equity / first loss
₹5 crore
Residual cash; absorbs losses first
“Senior” does not automatically mean AAA, and “equity” does not promise a high return. A waterfall sets the order in which available cash is used. A simplified interest waterfall: taxes and essential expenses → servicing and trustee fees → senior interest → mezzanine interest → reserve top-up → residual to equity. Principal may follow a separate waterfall.
The ₹3.9 crore is not guaranteed equity income: credit losses, reserve requirements and performance triggers can reduce or redirect it.
13 / LOSSES
Loss allocation, attachment and detachment points
Calculate losses after recoveries, then allocate them from the bottom of the structure. With defaults of ₹20 crore and recoveries of ₹8 crore, net loss is ₹12 crore: equity loses its full ₹5 crore, mezzanine loses ₹7 crore (₹8 crore remains), and the senior tranche is untouched.
Total pool loss
Equity loss
Mezzanine loss
Senior loss
₹3 crore
₹3 crore
Nil
Nil
₹12 crore
₹5 crore
₹7 crore
Nil
₹20 crore
₹5 crore
₹15 crore
Nil
₹25 crore
₹5 crore
₹15 crore
₹5 crore
The senior tranche starts losing principal only when pool losses exceed ₹20 crore, the protection beneath it. That protection level is its attachment point; attachment plus its own size is its detachment point.
Loss absorbed by a tranche = MIN(tranche par, MAX(0, pool loss − protection below it)) Attachment = share of the pool below the tranche · Detachment = attachment + tranche size
₹500 crore pool · 20% gross defaults × (1 − 40% recovery) = 12% net loss = ₹60 crore.
In the ₹500 crore structure above, a ₹60 crore loss wipes out the ₹35 crore equity and takes ₹25 crore from Class D, leaving it with ₹15 crore (37.5% intact). Classes A, B and C are untouched. These illustrations assume no excess spread, interest shortfall or other structural features.
14 / ENHANCEMENT
Credit enhancement: subordination, overcollateralisation, excess spread and reserves
Method
How it protects
Example
Subordination
Junior classes absorb losses first
₹20 crore of junior notes beneath the senior class
Overcollateralisation (OC)
Collateral exceeds the notes issued
₹1,000 crore of loans backing ₹955 crore of notes
Excess spread
Asset income above coupons and fees covers losses
Retained before any residual is released
Cash reserve
Cash set aside for specified shortfalls
A ₹15 crore reserve account
Guarantee / insurance
Third party covers defined obligations
Cover for specified payments
Auto ABS capital structure (₹1,000 crore pool, ₹955 crore of notes). Senior classes A-1, A-2 and A-3 are time-tranched: they are repaid one after another but share losses, so they share the same credit support.
Class
Size
Subordination below
OC
Hard credit support
A (A-1, A-2, A-3)
₹750 crore
20.5%
4.5%
25.0%
B
₹80 crore
12.5%
4.5%
17.0%
C
₹60 crore
6.5%
4.5%
11.0%
D
₹65 crore
0%
4.5%
4.5%
Class A can absorb pool losses of up to 25% before losing principal. A 1.5% reserve and roughly 3.8% a year of excess spread add soft support, which can be used up: do not add them to hard support as if they were permanent.
15 / CASE STUDY
Case study: Apex Residential Mortgage Trust 2026-1
Hypothetical teaching case. Apex Financial Corp. and the trust are invented names, and amounts are in US dollars. The case ties together every idea so far: pooling, true sale, tranching, the interest and principal waterfalls, and loss allocation.
A commercial bank, Apex Financial Corp., has originated $100 million of 30-year fixed-rate prime residential mortgages. Keeping them on its balance sheet ties up capital and leaves the bank exposed to interest-rate and liquidity risk, so Apex securitises the portfolio to recycle its capital.
Step 1 · Pool the loans and isolate them in an SPV
Pool characteristic
Value
Loans pooled
400 prime residential mortgages
Total balance
$100,000,000
Weighted average coupon (WAC)
6.50% a year
Weighted average maturity (WAM)
30 years (360 months)
Servicing and trustee fee
0.50% a year, kept by Apex as servicer
Net pass-through rate
6.00% a year: $6,000,000 in year 1
Deal structure: the true sale moves the loans into the SPV; the notes are issued by the SPV, not by Apex.
Step 2 · Capital structure and tranching
Tranche
Par value
% of pool
Rating
Coupon / return
Credit support
Senior, Class A
$80,000,000
80.0%
AAA
5.00% fixed
20.0% subordination
Mezzanine, Class B
$15,000,000
15.0%
BBB
6.50% fixed
5.0% subordination
Equity / residual, Class C
$5,000,000
5.0%
Unrated
Residual cash
None: first loss
Total issuance
$100,000,000
100.0%
The same structure seen two ways: a priority order for cash and a reverse order for losses.
Step 3 · The year-1 interest waterfall
In year 1 borrowers pay $6,500,000 of gross interest and $7,000,000 of principal (scheduled amortisation plus prepayments). For simplicity, interest in this case is calculated on the opening balances.
Bar lengths are drawn to scale: every dollar of the net interest is accounted for once.
Fee = $100M × 0.50% = $0.500M → net interest = $6.500M − $0.500M = $6.000M Class A interest = $80M × 5.00% = $4.000M (paid first) Class B interest = $15M × 6.50% = $0.975M (paid second) Class C residual = $6.000M − $4.000M − $0.975M = $1.025M → $1.025M ÷ $5M = 20.5%
The 20.5% equity return exists because the pool earns 6.0% net while the notes cost a blended (80 × 5.00% + 15 × 6.50%) ÷ 95 = 5.24%. That gap, the excess spread, is earned on $100M of collateral but paid to only $5M of equity: leverage of 20 to 1. It is a cash yield in a year with no losses, not a promised return.
Step 4 · The principal waterfall (sequential pay)
The whole $7,000,000 of principal goes to Class A, cutting it from $80M to $73M. Class B and Class C are unchanged, so the protection beneath the senior class rises from 20.0% to 20 ÷ 93 = 21.5%.
Sequential pay deleverages the senior class: the same $20M of junior notes now protects a smaller senior balance.
Step 5 · Stress scenario and loss allocation
Now suppose severe stress also causes $6,000,000 of loan defaults with a 50% recovery. The $3M recovered is collected as principal; the other $3M is a net loss, allocated in reverse order of seniority.
Recoveries repay the senior class like any other principal; the net loss writes down the equity.
Class C absorbs the whole $3M, falling from $5M to $2M. Classes A and B lose nothing: the equity buffer insulated them completely. Two refinements make the picture more realistic:
Excess spread first. Many deals use the year’s $1.025M of excess spread to cover losses before releasing anything to equity. Then the net write-down would be only $3M − $1.025M = $1.975M, and the equity cash payment would fall to zero.
Where the next loss goes. With only $2M of equity left, another net loss above $2M reaches Class B, and the senior class is hit only after Class B’s $15M is also gone, a cumulative net loss of $20M.
Try it yourself. The Securitisation Waterfall Simulator runs this kind of structure month by month over the full life of the pool; set Class A to 80%, Class B to 15% and raise the default rate to see when each class is reached. The Apex Case Study sheet in the Excel companion reproduces every number above with live formulas.
16 / CDOs AND CLOs
CDOs and CLOs: structure, equity returns and coverage tests
A CDO pools debt exposures and allocates their cash flows and losses across tranches. The portfolio may hold corporate bonds, corporate loans, ABS or MBS tranches, or other eligible debt.
Structure
Main collateral
CLO: collateralised loan obligation
Corporate loans, often leveraged loans
CBO: collateralised bond obligation
Bonds
Structured-finance CDO
ABS, MBS or other structured securities
A CLO of corporate loans and a CDO of mortgage-security tranches share the word “tranche”, not the economics.
How a CDO generates equity returns. On a ₹100 crore loan portfolio earning 11%:
This ignores defaults and principal changes: it is a simplified cash yield, not an IRR or a promised return. Equity earns the residual because it carries the first loss.
Coverage tests. Managed CLOs test their collateral every period. The OC ratio divides adjusted collateral par by the par of a class and every class above it. The IC ratio (interest coverage ratio, ICR) divides interest collected after senior fees and expenses by the interest due on that class and every class above it. A failed test diverts cash from junior classes to repay the senior notes. On a ₹500 crore portfolio with an ₹8.5 crore haircut on weak assets (adjusted par ₹491.5 crore) and ₹11.25 crore of quarterly interest available after senior fees:
Class (cumulative par)
OC ratio
Minimum
IC ratio
Minimum
A-1 (₹310 crore)
1.585
1.220
2.286
1.25
A-2 (₹360 crore)
1.365
1.150
1.947
1.20
B (₹390 crore)
1.260
1.100
1.778
1.15
C (₹420 crore)
1.170
1.065
1.617
1.10
D (₹445 crore)
1.104
1.035
1.462
1.05
Class
Par
Coupon (5% + spread)
Quarterly interest
Cumulative interest due
IC ratio
A-1
₹310 crore
6.35%
4.9213
4.9213
2.286
A-2
₹50 crore
6.85%
0.8563
5.7775
1.947
B
₹30 crore
7.35%
0.5513
6.3288
1.778
C
₹30 crore
8.40%
0.6300
6.9588
1.617
D
₹25 crore
11.75%
0.7344
7.6931
1.462
Class A-1 interest = 310 × 6.35% ÷ 4 = ₹4.9213 crore → IC(A-1) = 11.25 ÷ 4.9213 = 2.286
Class A-2 adds 50 × 6.85% ÷ 4 = ₹0.8563 crore → IC(A-2) = 11.25 ÷ (4.9213 + 0.8563) = 1.947
Each ratio uses the cumulative interest of the class and all classes above it, so the ratios fall as you move down the structure.
If downgrades raise the haircut to ₹45 crore, adjusted par falls to ₹455 crore and Class D’s OC ratio drops to 455 ÷ 445 = 1.022, below 1.035: the test fails and cash that would have reached Class D and equity is used to pay down the senior notes until the test is cured.
CLO life cycle. A warehouse phase assembles loans before closing; a ramp-up completes the portfolio; a reinvestment period (often several years) lets the manager recycle principal; an amortisation phase then repays notes sequentially from the top; equity may call the deal once a non-call period ends.
17 / CASE STUDY
Case study: Apex Global Corporate CDO I
Hypothetical teaching case. Apex Capital Management and the CDO are invented names; amounts are in US dollars. Compare it with the Apex RMBS case: the mechanics are the same, but the collateral, the leverage and the role of the manager change the risk.
A residential MBS bundles one kind of loan: home mortgages. A CDO pools a diversified mix of credit instruments. Here Apex Capital Management acts as the CDO’s asset manager: it assembles a $200 million high-yield corporate debt portfolio and funds it with cheaper senior debt, aiming to earn the spread between what the assets yield and what the notes cost: an arbitrage CDO.
Step 1 · Assemble the collateral and move it into the SPV
A diversified asset pool on the left, a three-layer liability structure on the right.
Step 2 · Tranching the liabilities
Class
Nominal
% of pool
Rating
Coupon
Subordination below
Senior, Class A
$140,000,000
70.0%
AAA
5.50% fixed
30.0%
Mezzanine, Class B
$40,000,000
20.0%
BBB
8.00% fixed
10.0%
Equity / first loss
$20,000,000
10.0%
Unrated
Residual cash
0%
Total
$200,000,000
100.0%
The CDO carries more subordination than the RMBS (30% under the senior class rather than 20%, and a 10% equity layer rather than 5%) because corporate high-yield credit is riskier and its defaults cluster in recessions.
Step 3 · Year-1 interest waterfall
Drawn to scale at 28 pixels per $1 million.
Fee = $200M × 0.75% = $1.5M → net interest = $17.0M − $1.5M = $15.5M Class A = $140M × 5.50% = $7.7M · Class B = $40M × 8.00% = $3.2M Equity residual = 15.5 − 7.7 − 3.2 = $4.6M → $4.6M ÷ $20M = 23.0%
Where does the 23% come from? The arbitrage, decomposed
The notes cost a blended (7.7 + 3.2) ÷ 180 = 6.06%, while the assets yield 7.75% net. Equity earns the net yield on its own $20M plus the spread on the $180M it has borrowed:
The second term is leverage at work: 9× notes per dollar of equity. It also explains the downside. If the asset yield falls, or defaults stop interest on part of the pool, that same 9× multiplier works in reverse.
Step 4 · Stress: corporate defaults
An economic shock causes $25M of defaults with a 40% recovery: $10M is recovered and $15M is a net principal write-down. (Equivalently, a $15M write-down at 40% recovery requires $15M ÷ 0.6 = $25M of defaulted loans.) The loss enters at the bottom.
Classes A and B lose nothing, but the cushion under Class B is now only $5M.
Gross defaults
Recovery
Net loss
Equity loss
Class B loss
Class A loss
Equity left
$5M
40%
$3M
$3M
$0
$0
$17M
$15M
40%
$9M
$9M
$0
$0
$11M
$25M (the case)
40%
$15M
$15M
$0
$0
$5M
$30M
40%
$18M
$18M
$0
$0
$2M
$50M
30%
$35M
$20M
$15M
$0
$0
$100M
40%
$60M
$20M
$40M
$0
$0
$120M
40%
$72M
$20M
$40M
$12M
$0
Read “a $15M write-down” carefully: a $15M gross default at 40% recovery is only a $9M loss (second row). Always convert gross defaults to net loss before allocating.
Step 5 · The year after: why equity cash does not just “drop proportionately”
Suppose the $10M recovered is used to repay Class A (the deal is past its reinvestment period). In year 2 the performing collateral is $175M and Class A is $130M:
Gross interest = $175M × 8.50% = $14.875M · fee = $175M × 0.75% = $1.3125M · net = $13.5625M Class A = $130M × 5.50% = $7.15M · Class B = $40M × 8.00% = $3.20M · residual before tests = $3.2125M
That residual is 16.1% on the original $20M of equity, or 64% on the $5M that remains. In a managed cash-flow CDO, though, an overcollateralisation test stands in the way:
OC test
Before the shock
After the shock
Illustrative minimum
Result
Class A: collateral ÷ A
200 ÷ 140 = 1.429
175 ÷ 130 = 1.346
1.220
Pass
Class B: collateral ÷ (A + B)
200 ÷ 180 = 1.111
175 ÷ 170 = 1.029
1.065
Fail
To cure the Class B test, Class A must be repaid until 175 ÷ (170 − x) ≥ 1.065, that is by x = $5.68M. The year’s whole $3.21M residual is therefore diverted to Class A, and equity receives nothing until the test passes again. The coverage test turns a 75% capital loss into a cash shut-off as well. This is the mechanism, described in Section 16, that protects the senior notes.
Try it yourself. In the simulator’s loss-waterfall lab, choose the Apex CDO I preset, then raise defaults to see where Class B and Class A are reached. The Apex CDO Case sheet in the Excel companion rebuilds the waterfall, the return decomposition, the stress table and the OC test with live formulas.
18 / CORRELATION
Default correlation: why expected loss is not enough
Default correlation measures how far borrowers tend to default together. A portfolio can hold many borrowers yet stay vulnerable if they depend on the same economy: a pool of property developers in one city can default together in a local downturn. For senior investors, clustered defaults can exhaust the junior protection.
Take 100 loans of ₹1 crore, each with a 10% default probability and zero recovery, a ₹80 crore senior tranche and ₹20 crore of junior protection.
Case A: independent defaults
Case B: perfectly correlated defaults
Expected pool loss
100 × ₹1 crore × 10% = ₹10 crore
90% × ₹0 + 10% × ₹100 crore = ₹10 crore
Chance the senior tranche loses anything
0.08% (more than 20 defaults)
10% (everything defaults)
Expected senior loss
about ₹0.0013 crore (≈ ₹1.3 lakh)
10% × ₹80 crore = ₹8 crore
Case A: defaults bunch around 10, and more than 20 is very rare. Case B has only two outcomes: none, or all 100.
The lesson: the same expected pool loss can hide very different tranche risk. It is the distribution of losses, especially severe clustered losses, that determines senior-tranche risk. The simulator’s correlation lab uses a one-factor (Vasicek) model to show this for any correlation between the two extremes.
19 / CRISIS
Synthetic CDOs, CDO-squared and the 2007–09 crisis
A synthetic CDO transfers credit risk through derivatives or guarantees, commonly credit default swaps, without necessarily buying the loans or bonds. A protection-selling investor earns premiums but bears losses after defined credit events in a reference portfolio, for example losses within a specified layer of ₹100 crore of corporate debt. The loans can stay with their original holders, which is why the Basel framework distinguishes synthetic from traditional securitisation.
A CDO-squared is a CDO backed by tranches of other CDOs: a new vehicle buys mezzanine tranches from several CDOs and issues new senior, mezzanine and equity classes. If those CDOs share similar underlying exposures, the apparent diversification is an illusion.
Several risks reinforced one another in the mortgage-related CDOs of the global financial crisis: weak mortgage underwriting, common exposure to the housing market, underestimated default correlation, resecuritisation of mortgage-security tranches, heavy reliance on ratings and models, and leverage that depended on market liquidity. When mortgage losses rose together, junior protection proved insufficient in some structures and losses reached classes regarded as highly protected.
Dimension
Pre-crisis mortgage-related CDO
Typical post-crisis CLO
Collateral
Often mezzanine RMBS tranches: a securitisation of securitisations
Senior secured loans to operating companies
Correlation
Housing defaults assumed to be weakly linked across regions
Correlated corporate defaults stressed explicitly
Leverage source
Synthetic exposure via CDS and short-term funding
Mostly cash-funded, term-matched notes
Safeguards
Thin protection relative to correlated losses
OC/IC tests, risk retention, more subordination
20 / VALUATION
Valuing MBS, ABS and CDO tranches
P = Σ E(CFt) ÷ (1 + rt)t E(CFt) = expected cash flow to this security · rt = discount rate for that timing and risk
Expected cash flows depend on defaults and recoveries, prepayments, fees, payment priorities, performance triggers and interest rates. A security expected to pay ₹8 at year 1 and ₹108 at year 2, discounted at 10%:
For mortgage securities the cash flows themselves change across interest-rate and prepayment scenarios, so one fixed schedule can mislead. That is why practitioners value them across many simulated paths and quote an OAS.
21 / DUE DILIGENCE AND INDIA
What to examine before investing, and the Indian market
Area
Questions to ask
Borrower quality
Who owes the money, and how strong is repayment capacity?
Collateral
What supports recoveries after default?
Concentration
Are exposures linked by geography, employer, industry or vintage?
Repayment behaviour
How sensitive are cash flows to prepayment?
Structure
Who is paid first and who loses first?
Enhancement
How much protection exists, and can it erode?
Triggers
Can deterioration redirect cash away from junior classes?
Servicing
Can collections continue if the servicer fails?
Legal
Is the transfer enforceable and properly isolated?
Valuation and liquidity
Can the security be valued reliably and sold when needed?
India. Transactions typically securitise loan pools originated by banks, NBFCs and housing finance companies through a trust, which issues pass-through certificates (PTCs). A related route is direct assignment, where loans are transferred directly to a buyer rather than through securities.
Transaction
Basic mechanism
Securitisation
A pool supports the issue of securities (PTCs) by a trust
Direct assignment
Loans are transferred directly to a purchaser
The Reserve Bank of India’s Securitisation of Standard Assets Directions of 24 September 2021 set, among other things, a minimum retention requirement (5% of the book value of the loans securitised where their original maturity is up to 24 months; 10% where it is longer, or where loans have bullet repayments; 5% for residential mortgage-backed transactions irrespective of maturity), a minimum holding period (3 months for loans of up to two years’ tenor, 6 months otherwise) and a ₹1 crore minimum ticket size for securitisation notes. The applicable requirements depend on the regulated entity and transaction, so check the current version on the RBI website. International structures such as revolving pools, synthetic transactions or resecuritisation should not be assumed to be permitted in India.
22 / LAB
Interactive simulator and Excel companion
Securitisation Waterfall Simulator
Build a mortgage or loan pool, set prepayment (CPR or PSA) and default speeds, tranche it into senior, mezzanine and equity, and follow the monthly waterfall. The simulator writes losses down bottom-up and reports each tranche’s loss, WAL and yield. It also contains a loss-allocation lab, a prepayment and WAL lab, and a default-correlation lab that reproduces Section 18 for any correlation.
Try this: start with the base deal (senior 80%, mezzanine 12%, equity 8%), then choose Default stress. Cumulative net loss rises to about 29%: equity is wiped out and mezzanine is nearly lost, while the senior class survives on its 20% subordination plus excess spread.
Excel companion: thirteen formula-driven sheets
Apex RMBS case · Apex CDO case · pass-through and CPR ↔ SMM · 12-month PSA schedule with seasoning, gross WAC, WAM and CDR · WAL and sequential CMO · auto ABS cumulative net loss trigger · interest waterfall and CDO equity yield · loss allocation with attachment and detachment · auto ABS support and a card ABS trigger · CLO OC and IC tests · default correlation · valuation. Every example from this article is pre-filled; blue cells on yellow are inputs.
Borrower, market, operational and legal risks remain
Equity always earns the most
It has residual upside and takes the first loss
Tranching creates extra cash
It reallocates the pool’s cash flows and losses
SMM = 1 − (1 − CPR)1/12 · MDR = 1 − (1 − CDR)1/12 · 100% PSA: CPRt = 6% × MIN(loan age, 30) ÷ 30
Gross WAC = Σ(balance × rate) ÷ Σ balance · Net WAC = gross WAC − fees · WAM = Σ(balance × term) ÷ Σ balance
WAL = Σ(t × principalt) ÷ Σ principalt · Cumulative net loss = (defaults − recoveries) ÷ original pool
Excess spread = asset yield − weighted note coupon − fees · OC ratio = adjusted collateral ÷ (class par + all senior par)
IC ratio = (interest collected − senior fees) ÷ interest due on the class and all senior classes
Why can a faster prepayment speed hurt an MBS investor who bought above par?
Principal is returned at par, so a premium paid is lost sooner, and the cash must be reinvested at the lower yields that triggered the refinancing. This is contraction risk.
A ₹100 crore pool has 5% equity and 15% mezzanine. Gross defaults are ₹30 crore with 50% recovery. Who loses what?
Net loss = 30 × (1 − 0.5) = ₹15 crore. Equity loses ₹5 crore (wiped out), mezzanine loses ₹10 crore (₹5 crore left), senior loses nothing. Senior losses start only above ₹20 crore of net loss.
Convert a CPR of 20% to an SMM.
SMM = 1 − 0.801/12 = 1.842% a month.
Why do A-1, A-2 and A-3 notes in an auto ABS have the same credit support?
They are repaid in time order but rank equally for losses, so all three rely on the same subordination and overcollateralisation beneath Class A.
Two pools both have a ₹10 crore expected loss. Why can their senior tranches carry very different risk?
Expected loss is a mean. If defaults are independent, large losses are extremely unlikely; if they are highly correlated, the pool sometimes loses most of its value, which reaches the senior tranche. Tranche risk depends on the whole loss distribution.
What happens when a CLO’s Class D OC test fails?
Interest that would have gone to Class D and the equity is diverted to repay the senior notes (or buy collateral) until the ratio returns above its minimum. This protects senior noteholders at the expense of junior ones.
Two loan groups: ₹60 crore at 7% and ₹40 crore at 6%. Fees are 0.5%. What are the gross and net WAC?
Gross WAC = (60 × 7% + 40 × 6%) ÷ 100 = 6.6%. Net WAC = 6.6% − 0.5% = 6.1%. The simple average, 6.5%, is wrong because the groups differ in size.
A ₹200 crore auto pool has cumulative defaults of ₹9 crore and recoveries of ₹4 crore. What is the CNL, and does it breach a 2% trigger?
CNL = (9 − 4) ÷ 200 = 2.5% of the original pool. It is above 2%, so the trigger is breached and cash is redirected to protect the senior notes.
A CLO collects ₹12 crore of interest, pays ₹1 crore of senior fees and owes ₹5 crore of Class A interest. What is the Class A IC ratio?
IC = (12 − 1) ÷ 5 = 2.2×. Interest available after senior fees covers the Class A coupon 2.2 times.
24 / SOURCES
Sources and connected learning
Teaching notes by Prof. V. Ravichandran, with every numerical example independently recomputed. All amounts are hypothetical illustrations, not descriptions of actual transactions.