Mortgage Securitization, FinTech & Historical Computing Giants Lecture

Cash-Flow Discounting & Basic Deal Structure

• Future dollars are worth less today: This idea means that money you have in your hand right now is more valuable than the same amount of money you expect to receive in the future. The reasons for this are that you could invest money today to earn more, or inflation might make future money buy less.

• Buyer offers $0.95$ for a $1.00$ payment due in one year: Imagine someone owes you a dollar in a year. A buyer might offer to give you 95 cents for that dollar today. They pay less because they have to wait to get the full dollar.

• Implied simple-interest return fisher approx $r = \frac{1-0.95}{0.95} \approx 5.26\%$: This formula shows the approximate percentage profit, or interest rate, you would earn. If you bought that future dollar for 95 cents today, your profit (return) would be about 5.26%.

• Bank originates a mortgage (30-yr, 10-yr etc.): When a bank "originates" a mortgage, it means they are giving out a brand new home loan to someone. These loans are typically paid back over a long period, like 30 years.

• Now owns a stream of future cash-flows (principal + interest): Once the bank issues a mortgage, they are then owed many regular payments from the borrower over many years. These payments include the original amount borrowed (the "principal") plus extra money as a fee for borrowing (the "interest").

• Instead of waiting decades, the bank sells that stream at a discount: Banks often don't want to wait 30 years to get all their money back from a mortgage. So, they sell the right to receive these future payments to other investors. They sell it for less than the total amount of all future payments (at a "discount") to get cash right away.

• Raise immediate cash and recycle capital into new loans: By selling these mortgages, banks get cash quickly. This allows them to use that money to give out even more new loans, which helps more people buy homes and keeps the economy going.

• Major bank revenue sources:

  • Up-front transaction fees ("points", application costs): These are one-time fees that borrowers pay to the bank when they first get a mortgage. "Points" are expressed as a percentage of the loan amount, for example, 1 point equals 1% of the loan.

  • Gain-on-sale when mortgage is sold to highest bidder: When a bank sells a mortgage to another investor, it aims to sell it for more than it cost the bank to create the loan. This profit is called a "gain-on-sale."

Motivation of Key Players

• Borrower / Home-buyer: This is the individual who takes out the mortgage loan to purchase a house. They commit to repaying the loan over time, usually through monthly payments.

• Originating Bank: This is the bank that first issues the mortgage loan. They make money from fees and by selling the mortgage. Selling mortgages "frees balance-sheet capacity," which means it clears up space on their financial records, allowing them to lend even more money without depleting their funds to lend.

• Mortgage Buyer / Aggregator (e.g., Goldman, Citi, Fannie, Freddie): These are companies or organizations that buy a large number of mortgages from various banks. Their primary goal is to gather thousands of these mortgages and then either resell them or create new investment products from the future payments, aiming to make more money.

• Institutional Investors (insurance cos., pension funds, mutual funds): These are large groups that manage money for many people (like retirement savings or insurance). When they invest in mortgages, they "seek steady coupons," meaning they want regular, predictable interest payments. They are looking for a "slightly higher yield than Treasuries," which are very safe US government bonds, meaning they want a bit more profit for taking on a bit more risk.

• Key economic claim: when structured well, the chain is not a zero-sum game: This means that when the system is set up correctly, everyone involved – the borrower, the bank, those who buy the mortgages, and the investors – can benefit. It's not a situation where one person's gain necessarily means another's loss. Instead, by making it easier for money to move around, everyone can potentially achieve their financial goals, though some might gain more than others.

Securitization Mechanics

• Process name: Securitization: This is the process of taking individual loans that are hard to sell on their own (like a single mortgage) and combining them into new, easy-to-sell financial products, similar to bonds, that can be traded in financial markets.

• Steps:

  1. Aggregator pools $\approx10^3$ mortgages (~$\$1\text{ billion}$ notional): The company that buys mortgages (the "aggregator") gathers a large number of them, perhaps around 1,000. The "notional" value refers to the total original loan amount, which might be about one billion dollars.

  2. Creates bonds backed by the pool: The aggregator then uses this large group of mortgages as collateral or a promise to support new investment products, which are essentially bonds. These bonds promise to pay investors money that comes from the payments made on the original mortgages.

  3. Splits bonds into tranches with different rights: These new bonds are then divided into different "slices" or "tranches." Each slice has different rules about when and how its investors get paid, which affects how much risk they take and how much money they can potentially earn.

• Advantages:

  • Diversification: Instead of owning just one mortgage (which is risky if that particular borrower defaults), an investor can buy a security that is supported by thousands of mortgages. Since these loans come from many different people and places, the risk is spread out. If one person can't pay, it's a very tiny problem within the huge group of loans.

  • Smaller bite-sizes: You can usually buy a piece of these investments for around $\$1,000$ or similar small amounts. This makes them affordable and accessible to more types of investors, not just very wealthy people or large financial institutions.

  • Frees up bank capital $\rightarrow$ more lending $\rightarrow$ economic expansion: When banks sell mortgages, they get back the money that was tied up in those loans. They can then use this freed-up money ("capital") to give out even more new loans. This increased lending helps the economy grow by making more money available for homes, businesses, and other investments.

Pass-Throughs vs. CMOs

• Pass-Through (PT):

  • Simplest form: This is the most basic type of investment created from mortgages.

  • Monthly borrower payments "pass through" pro-rata to investors: When people make their monthly mortgage payments (both principal and interest), after a small fee for managing the loan, the rest of the money goes directly to the investors. Each investor receives a share ("pro-rata" means proportional) based on how much of the pass-through security they own.

  • Cash-flow = monthly coupon + eventual principal return: Investors in a pass-through security receive regular monthly payments that are like interest (called a "coupon") plus a portion of the original loan amount being paid back. Eventually, they get all of their original investment back.

• Collateralized Mortgage Obligation (CMO):

  • Cash-flows carved into tranches with tailored timing/priority: Unlike simple pass-throughs, CMOs take the incoming mortgage payments and divide them into several different "slices" (tranches), each of which pays out money differently. For example, some slices might get their main loan amount paid back first, while others might only receive interest payments for a long time. This is done to create specific payment schedules for different types of investors.

  • Possible bespoke structures: CMOs can be custom-made to fit the very specific needs of investors. For instance, a slice could be designed to pay out mostly in certain years, like when an investor needs money to pay for their child's college tuition.

  • Offers targeted duration, convexity, and risk/return profiles: Investors can choose slices that perfectly match their investment goals. "Duration" is a measure of how much a bond's price changes when interest rates change. "Convexity" is a more complex measure related to that change. CMOs allow investors to pick the exact level of interest rate risk and the expected profit they want.

• Pricing depends on Option Adjusted Spread (OAS): OAS is a special measurement used to determine the fair price of complex bonds like CMOs. It shows the extra profit (the "spread") an investor gets compared to a very safe investment (like government bonds), after considering any "embedded options." The main embedded option in a mortgage is that the borrower can choose to pay back their loan early (pre-pay). OAS is calculated using complex financial computer models.

Pre-Payment & Embedded Options

• Two drivers of pre-payments (both terminate cash-flow): "Pre-payment" means a borrower pays off their mortgage loan sooner than originally planned. When this happens, the regular monthly payments to the investor stop (the "cash-flow terminates").

  1. Refinance when market rates $\downarrow$: If interest rates generally go down, borrowers might choose to get a new loan with a lower interest rate to pay off their old, more expensive mortgage. This is a common reason for pre-paying.

  2. Default / inability to pay: If a borrower can no longer make payments at all, the loan is considered defaulted, which also stops the expected payments.

• Aggregator/investor faces call-option-like risk: When you own a mortgage (or a security backed by mortgages), it's somewhat like you've given the borrower a "call option." This means the borrower has the right (but not the obligation) to pay off their mortgage early. This is a risk for the investor because pre-payments usually happen when it's financially unfavorable for the investor (for example, when interest rates drop, meaning the investor has to reinvest the money they get back at a lower rate). This is why OAS modeling is crucial to understand and price this risk.

• YieldBook innovation: embedded mortgage model to project pre-payments $\rightarrow$ calculate OAS quickly on hundreds of bonds overnight: Before a tool called YieldBook, figuring out how likely mortgages were to pre-pay and calculating their OAS was a very slow process. YieldBook changed this by putting a smart mathematical model right into its software, allowing users to quickly estimate pre-payment rates and get instant OAS calculations for many bonds.

YieldBook & FinTech Disruption

• Old workflow: In the past, if an investor wanted to know detailed information about a mortgage bond, they would typically ask a sales person, who would then ask a research analyst. The crucial OAS number might arrive hours later, or even the next day, often by fax.

• YieldBook placed model on user desktop; click Calculate $\rightarrow$ instant OAS: YieldBook completely changed this by putting the complex math model directly on the user's computer. This meant investors could get real-time OAS calculations instantly by just clicking a button.

  • Removed middle-men, reduced latency, democratized analytics: It eliminated the need for people in between (sales representatives, analysts) and greatly reduced the delay ("latency") in getting information. This made advanced financial analysis available to many more people ("democratized analytics"), not just a select few.

  • Rapid adoption: It quickly went from being used by only a few people to being adopted by hundreds of financial companies and thousands of individual users.

  • Sparked larger trading volumes by increasing transparency: Because analysis was faster and clearer, investors felt more confident trading mortgage-backed securities, leading to more buying and selling of these products.

Peer-to-Peer Case Study – LendingClub

• Connects Debtors (high-rate credit-card holders) with Lenders via web platform: LendingClub is an online platform that directly connects people who need to borrow money (often "debtors" trying to pay off expensive credit card debt with high interest rates) with individuals or groups who want to lend money ("lenders").

• Typical numbers: For example, someone might be paying 25% interest on their credit card. LendingClub might offer them a loan at 11%. LendingClub keeps 2% as its fee for arranging the loan, and the lenders earn 9% interest on their money.

• Key FinTech elements:

  • Proprietary data-driven credit-worthiness model (beyond FICO): LendingClub uses its own smart computer programs and data analysis to figure out how risky it is to lend money to a borrower. This model goes deeper than just using a traditional credit score like FICO, allowing them to make more precise lending decisions.

  • Disintermediation of banks: "Disintermediation" means removing the middleman. In this case, LendingClub takes traditional banks out of the process of giving smaller loans (like $\$20k$), because it's often too expensive and difficult for large banks to manage such small loans profitably.

• Parallel to mortgage chain: Just like with mortgages, LendingClub's platform collects loan payments from borrowers and sends them to the lenders, taking a small fee for managing everything.

Investor Risk Tolerance & Institutions

• Insurance Cos.: These companies are closely regulated because they need to be prepared to pay out huge amounts of money (for instance, after a natural disaster). They require "high-quality, liquid bonds" so they can easily sell them for cash quickly ("liquid") to handle large "catastrophe payouts" if needed.

• Pension Funds: These funds hold money for people's retirement. How much they expect to pay out to retirees over time (their "liability profile") is quite steady and can be predicted using "actuarial life tables" (which are statistics about how long people live and for how long they will receive payments). This means they can accept "moderate duration risk," meaning they are okay with some ups and downs in their bond values due to interest rate changes, because their investments are for the very long term.

• Hedge Funds: These are investment funds that use more aggressive and sometimes riskier strategies. They often invest in "higher-risk CMOs" (more complicated and volatile parts of mortgage bonds) or "derivatives" (financial contracts whose value comes from something else, like a stock or bond) to try and achieve bigger profits.

• Personal advisory: When a financial advisor helps individuals, it's very important to understand their "loss tolerance" – how much money they can afford to lose without feeling too uncomfortable or facing significant financial hardship. An advisor might ask, "How would you feel if your investments dropped by 30% (30% drawdown)?" to gauge their comfort level with risk and determine suitable investments.

Technology Curve & Impact on Finance

• Cost-of-compute graph: Flops per $\$1$ has grown from $10^{-4}$ (1937) to $10^{12} \text{ + today}$: This graph illustrates how incredibly powerful and cheap computers have become. "FLOPs" (Floating Point Operations Per Second) measure a computer's speed and ability to perform calculations. This means that for just one dollar, you can now get trillions of calculations done, compared to almost none in the past. This makes computing extremely powerful and affordable for widespread use.

• Ubiquitous super-computing $\rightarrow$ affordable AI, big-data mortgage modelling (Michael Burry’s loan-level analysis in "The Big Short"): Because powerful computing is now everywhere and inexpensive, advanced technologies like Artificial Intelligence (AI) and complex analysis of huge amounts of data ("big-data") have become affordable. This allowed investors like Michael Burry, famously depicted in "The Big Short" movie, to examine individual mortgage loans very closely ("loan-level analysis") to discover hidden risks within them.

• Warning: current Gen-AI often polite but shallow: Modern AI tools (like chatbots) can give answers that sound good and polite, but they might not truly understand the topic in depth or provide profound insights. It's important to "probe beyond surface answers," meaning you should dig deeper and not just accept what the AI says at first glance, but critically evaluate it.

Historical Giants Highlighted

• Ada Lovelace (1815-1852): She is often recognized as the world's first computer programmer. She realized that Charles Babbage's Analytical Engine (an early mechanical computer) could do much more than just crunch numbers; she envisioned it creating things like music and art. She wrote detailed steps (similar to "pseudo-code") for how this machine could perform complex tasks.

• David Hilbert (1862-1943): A famous mathematician who proposed 23 very difficult problems in mathematics. One of these problems led to the significant question of "decidability" (whether a problem can be solved by a step-by-step process, or algorithm) and the "halting problem" (whether an algorithm will ever finish running).

• Alan Turing (1912-1954): A brilliant mathematician and computer scientist. He developed the idea of a "Universal Machine" (a theoretical computer model that can simulate any computer algorithm) and proved that the halting problem is "undecidable" (meaning no algorithm can determine if another arbitrary algorithm will halt or run forever). He was instrumental in breaking the Enigma code during World War II (as dramatized in "The Imitation Game" movie) and created the "Turing Test" to evaluate if a machine can exhibit intelligent behavior equivalent to, or indistinguishable from, a human.

• Grace Hopper (1906-1992): A pioneering U.S. Navy Rear Admiral and computer scientist. She created the first "compiler" (a program that translates human-readable programming code into machine code that a computer can understand and execute) for COBOL, an early business-oriented programming language. She also popularized the term "computer bug" after finding a real moth stuck in a computer system causing a problem.

• Donald Knuth (1938-): A very important computer scientist, renowned for writing the multi-part book series "The Art of Computer Programming," which covers fundamental algorithms and data structures. He also created the TeX/LaTeX system, which is widely used for precisely typesetting complex mathematical and scientific documents.

Key Formulas & Quantitative References

• Present Value: $PV = \frac{CF<em>t}{(1+r)^t}$: This formula calculates the "present value" (PV), which is how much a future amount of money (CFt, meaning cash flow at time t) is worth today. It takes into account how much you could earn on that money (r, the discount rate) and how long you have to wait for the money (t, the number of time periods).

• Basic Bond Cash-Flow (monthly): $CF = P \times i<em>m + \text{scheduled principal}$: This formula shows the amount of money you receive each month (CF) from a bond. It's calculated by multiplying the remaining loan amount or principal (P) by the monthly interest rate (im), and then adding any part of the original loan amount that is scheduled to be paid back that month.

• Option Adjusted Spread: $OAS = y<em>{\text{bond}} - y</em>{\text{benchmark}} - \text{option cost}$ (solved via model): As explained earlier, OAS is a complex way to measure a bond's potential profit. It represents the total profit the bond is expected to yield ($y<em>{\text{bond}}$), minus the yield from a very safe investment (the $y</em>{\text{benchmark}}$ or comparable government bond), and then further subtracts the estimated cost or value of any options the borrower has (like the choice to pay early). This calculation typically requires a sophisticated financial computer model to determine.

Ethical & Practical Implications

• Securitization can broaden home ownership & economic growth: By making it easier for banks to lend money, securitization can help more people get loans and buy homes. This increased access to housing and credit can then boost the overall economy. Yet mis-rating or lax underwriting (2008) creates systemic risk: However, if the underlying loans are poorly judged for their risk ("mis-rating") or given out too easily without sufficient checks ("lax underwriting"), it can lead to many people being unable to pay their loans, as happened in the 2008 financial crisis. This creates "systemic risk," meaning a risk that could potentially cause the entire financial system to become unstable or collapse.

• FinTech democratizes access (LendingClub, micro-loans): Financial technology ("FinTech") uses technology to make financial services available to more people, especially those who might not get services from traditional banks. Examples include online platforms like LendingClub or very small loans (micro-loans) to underserved populations. But must manage data privacy, fair-lending, model bias: However, FinTech companies must carefully protect users' private information, ensure their services are fair and do not discriminate ("fair-lending"), and ensure their computer models don't have unfair biases based on the data they use (e.g., unintentionally discriminating against certain groups).

• AI tools raise productivity yet may homogenize writing & obscure nuanced truths: Artificial Intelligence can make work much faster and more efficient ("raise productivity"). But relying too much on AI might make writing sound too similar and unoriginal ("homogenize writing") and could make it harder to find or express complex or subtle ideas ("obscure nuanced truths"). Therefore, "critical thinking remains indispensable" – it's still extremely important for people to analyze, evaluate, and truly understand information themselves, rather than just relying on AI-generated outputs.

Connections & Real-World Relevance

• Mortgage pool analytics foundational to fixed-income desks, risk management, regulatory capital: Understanding how to analyze large groups of mortgages is a fundamental and essential skill for many areas in finance: "fixed-income desks" (teams that trade bonds and debt securities), "risk management" (identifying, assessing, and reducing financial dangers within an institution), and calculating the "regulatory capital" (the amount of money banks must keep aside to cover potential losses) required by financial rules and regulations.

• Skills Blend: To succeed in this field, you need a mix of different skills: understanding financial concepts, using statistical math models, knowing how to program computers, and continuously learning because technology changes very rapidly.

• Personal action items suggested by lecturer:

  • Watch "The Big Short" & "Imitation Game" to see these concepts explained in a story format.

  • Check your FICO score to understand your own credit history and how it affects borrowing.

  • Look at a brokerage website to find MBS (Mortgage-Backed Security) listings to see how these investments are listed for sale in the market.

  • Practice using LaTeX for presenting mathematical formulas clearly, and read research papers about YieldBook and OAS to learn more deeply about their functions and calculations.