Foundations of FinTech · FIN 6778

Exam 1 Study Lab

Understand the friction. Follow the incentives. Practice applying the concepts.

Lecture focus: Weeks 1, 2, 4, 5 and 6

Start here

Your study plan

This independent study aid draws from lectures before Exam 1, the two case discussions and the supplied practice questions. The practice questions are not a comprehensive syllabus; check course announcements for authoritative coverage.

01 · Weeks 1–2

Banking fundamentals

Financial intermediation, information problems, risk and bank capital.

02 · Week 4

Open banking

Unbundling, data sharing, competition and the JPMorgan case.

03 · Week 5

FinTech lending

Credit scoring, marketplace funding, incentives and crowdfunding.

04 · Week 6

Data & models

Alternative data, prediction, fair lending and the Upstart case.

Lecture review

Open a section for a concise review. Examples and questions are original study aids.

Abbreviations and terminology
  • API: application programming interface. ML: machine learning. AI: artificial intelligence.
  • ROA: return on assets. ROE: return on equity. MBS: mortgage-backed security.
  • FICO: the Fair Isaac Corporation credit-score brand. ECOA: Equal Credit Opportunity Act. FCRA: Fair Credit Reporting Act.
  • P2P: peer-to-peer. SEC: Securities and Exchange Commission. FDIC: Federal Deposit Insurance Corporation.
Weeks 1–2 · Functions of finance and FinTech
  • Finance channels funds from savers to borrowers and investments. A financial institution can reduce search and transaction costs, pool risk, evaluate borrowers and support payments. A strong financial system can support productive investment; an observed correlation between finance and growth does not by itself prove causation.
  • Debt generally promises contractual payments; equity is a residual ownership claim. An exchange is an organized trading venue; over-the-counter transactions occur outside an exchange.
  • Study FinTech through the friction it addresses: unit cost, information asymmetry, agency conflicts, trust, search and delays or inconvenience. Technology can reduce a friction yet create costs or new risks.
  • A fintech can unbundle a bank service, such as payments or loan origination, and later rebundle services. Legacy systems, switching costs, regulation and network effects shape competition. Regulatory arbitrage means taking advantage of differences in applicable rules, not necessarily breaking them.

Sources: Week 1 Traditional Finance and Introduction to FinTech slides; Week 2 Banking and FinTech slides; Weeks 1–2 lecture reports.

Weeks 1–2 · Information, incentives and bank risk
  • Adverse selection: hidden borrower type before lending. If a lender cannot distinguish risk, common terms can repel safer applicants and leave a riskier pool—the lemons problem. Screening addresses it.
  • Moral hazard: a borrower changes behavior after receiving funds, perhaps taking more risk. Monitoring, covenants and incentives can address it.
  • Agency problem: one party's incentives differ from another's. An originator paid for volume may screen less carefully when investors bear default losses. Reputation, oversight and risk retention can mitigate, not eliminate, the conflict.
  • Liquidity risk: inability to meet withdrawals or payments when due. Credit risk: borrower fails to pay. Interest-rate risk: rate changes affect income or asset values. Operational risk: failures of processes, systems or people. Liquidity and solvency are distinct, although one can worsen the other.
  • Banks typically fund long-term assets with shorter-term liabilities: maturity transformation creates value and exposes them to withdrawal and rate risks. Capital is an equity cushion against losses; cash reserves help meet near-term payments.

Sources: Week 1 Traditional Finance and Week 2 Banking and FinTech slides; supplied practice questions 2–3, 10–11 and 16.

Weeks 2–4 · Bank balance sheets and innovation
  • Balance sheet: assets = liabilities + equity. Loans and securities are bank assets; customer deposits are liabilities. Loan losses reduce asset value and, other things equal, equity.
  • ROA = net income ÷ assets. ROE = net income ÷ equity. Example: $1 million net income, $100 million assets and $10 million equity yield 1% ROA and 10% ROE. More equity provides a bigger loss cushion but lowers ROE if assets and income are unchanged.
  • Simple deposit multiplier: in a stylized closed system with a 10% required reserve ratio and no cash or excess-reserve leakage, $100 of new reserves supports at most $1,000 in total deposits, including the initial deposit. This is a classroom illustration, not a description of current U.S. reserve requirements or a guarantee of actual lending.
  • Financial innovation may follow changes in demand (interest-rate risk), supply (cheaper computing) or incentives to avoid existing restrictions. Examples include adjustable-rate loans, electronic payments and historical alternatives to deposit-rate ceilings.

Sources: Week 2 Banking and FinTech slides and Week 4 Banking and FinTech slides; supplied practice question 12.

Week 4 · Open banking and JPMorgan case
  • Open banking enables consumers to authorize third-party access to financial data. Potential benefits include account aggregation, switching and new services; privacy, security and competition concerns remain.
  • Screen scraping: a third party logs in with customer credentials and extracts information from the bank's interface. It can expose credentials and collect more information than needed.
  • Application programming interfaces (APIs) provide structured machine-to-machine access and can use permissioned tokens and limited data scopes. The specifics depend on design; an API does not by itself eliminate third-party risk. Banks can influence access through cooperation, standards and contractual terms.
  • In the historical JPMorgan case, the bank cited security and privacy concerns while fintechs raised competition and access concerns. Both can be relevant simultaneously. A constructive analysis weighs customer control, security, bank incentives, access scope and accountability.

Sources: Week 4 Banking and FinTech slides; JPMorgan Chase & Co.: Open Banking, case pp. 1–4; supplied practice questions 6–7 and 14.

Week 5 · Credit scores and alternative information
  • Credit helps finance purchases and manage timing gaps, but borrowers may lack credit history, collateral or a guarantor. A credit-invisible person has no traditional credit file; a thin or unscorable file contains insufficient information for a given score.
  • Common FICO score categories include payment history, amounts owed, length of history, new credit and credit mix. Utilization = revolving balance ÷ credit limit; $600 on a $2,000 limit is 30%. The exact scoring formula is proprietary and depends on the score version.
  • Alternative data is relative to the lender and use: rent, utilities, employment or transaction records may add information beyond a traditional bureau file. More data does not automatically mean a better or fairer model.
  • Blended: use traditional and alternative data together in the primary assessment. Second look: use alternative information when an initial conventional assessment is missing or insufficient.

Sources: Week 5 and Week 6 FinTech Lending slides; supplied practice question 17.

Week 5 · Marketplace lending and crowdfunding
  • A marketplace lender typically matches borrowers with banks or investors and may earn origination, referral and servicing fees; who funds, originates, owns and services a loan depends on the arrangement. Investors or lenders holding loans face credit losses. A balance-sheet lender holds loans and bears the associated exposure.
  • A nonbank mortgage originator may use wholesale or investor funding and sell originated loans or mortgage-backed securities. Selling can replenish funding; it may weaken screening incentives if the originator lacks exposure, though contracts and reputation matter.
  • Marketplaces face credit, platform, legal and liquidity risks. Investors may not be able to sell an illiquid loan quickly; the platform also needs trust on both sides of the marketplace.
  • Reward-based crowdfunding: supporters receive a promised product or other non-equity benefit. Equity crowdfunding: investors receive an ownership stake with potential gains and losses. Neither guarantees success.

Sources: Week 5 FinTech Lending slides; Week 1 Traditional Finance slides; supplied practice questions 1, 15–16.

Week 6 · Data, machine learning and fairness
  • Structured data follows a defined schema, such as tabular loan fields. Unstructured data includes free text or images. Semi-structured data has partial organization, such as a filing with headings and free text.
  • Supervised learning trains using known outcome labels; predicting default from observed defaults is classification. Unsupervised learning identifies patterns without a labeled target, such as clustering applicants. Deep learning uses multilayer neural networks and is a subset of machine learning, not its synonym.
  • Overfitting: a model learns training-sample noise and performs poorly on new data. Select complexity using validation or cross-validation, then evaluate once on an untouched test set. The goal is out-of-sample performance, not the smallest training error.
  • Selection bias: repayment outcomes only for previously approved borrowers may not represent a broader applicant population. More rows alone do not fix this. Ranking performance does not itself demonstrate calibration, fairness or causality.
  • A variable can proxy for protected characteristics even if those characteristics are excluded. Assess data quality, privacy, legal obligations, group outcomes and possible less-discriminatory alternatives. Relevant credit rules can require specific adverse-action explanations; correlation alone does not establish a legal violation.

Sources: Week 6 FinTech Lending slides; Week 6 lecture report; supplied practice questions 4–5 and 18–19.

Week 6 case · Upstart and Alliant
  • The case's capital-light Upstart platform assessed and serviced loans and earned primarily referral, platform and servicing fees; partner banks and credit unions and institutional investors funded or held much of the lending. The roughly 5% balance-sheet share mentioned in the practice questions describes that case period, not a permanent rule.
  • Early case-period features included education and employment information and application interactions. The case describes subsequent removal of average SAT/ACT scores from school grouping. Predictive usefulness alone does not settle fairness or compliance.
  • In the case's FICO 660–679 band, the reported annualized default rates were 0.7% for Upstart's A+ grade and 8.0% for E-minus. The comparison suggests the grades contained additional risk information within that sample; it does not show grading caused repayment or prove fairness.
  • For Alliant, potential benefits included growth, technology and access to applicants missed by conventional scoring. Risks included fair-lending exposure, partner dependence, reputational harm and model performance under different economic conditions. Consider validation, explainability and ongoing oversight.

Sources: Week 6 Case Study 2 slides; Upstart's Upshot: Is Fintech Lending Fair? case; supplied practice questions 8–9 and 20. Case figures and facts are historical.

Source notes

The review summarizes course lecture slides for Weeks 1, 2, 4, 5 and 6, the assigned JPMorgan open-banking and Upstart case readings, and the supplied Exam 1 practice questions. Lecture reports informed the organization and interpretation. Case claims and historical examples are not presented as current company or regulatory facts. No source files or network access are needed to use this page.