DK-011 — Banking as a System
🏦 Banking as a System: Products, Risk, and Human Behavior
When people think about banks, they often think about products: accounts, loans, cards, apps.
But for anyone working inside a bank—especially as an AI researcher or developer—banking is not a collection of products.
Banking is a system.
A system where money, time, risk, regulation, and human behavior interact continuously.
This chapter explains:
- what banks actually sell
- how banking products fit together
- why risk management dominates everything
- why human behavior matters as much as mathematics
1️⃣ What a Bank Really Does
At its core, a bank performs three fundamental functions:
-
Liquidity transformation
Turning short-term money (deposits) into long-term money (loans) -
Risk transformation
Pooling, pricing, and redistributing risk across borrowers -
Trust intermediation
Convincing millions of people that their money is safe
Everything else—apps, branches, AI, products—exists to support these three goals.
2️⃣ Core Banking Products
2.1 Deposits: The Bank’s Raw Material
Deposits are not just customer products.
They are the input of the banking system.
Common types include:
- Savings accounts
- Current / checking accounts
- Fixed-term deposits
From a systems view:
- deposits are liabilities on the bank’s balance sheet
- but they are also the cheapest and most stable funding source
Longer-term deposits are especially valuable because they reduce liquidity and refinancing risk.
2.2 Loans: Where Risk Lives
Loans are where banks earn money—and where banks can fail.
Major categories:
- Retail loans (personal, housing, auto)
- SME and business loans
- Corporate and project finance
Each loan embeds assumptions about:
- future income
- economic conditions
- human behavior
This is why lending is not just a math problem. It is a prediction under uncertainty.
2.3 Savings-with-Incentives Products
Some banking products exist primarily to shape behavior.
Examples include:
- prize-linked savings
- lottery-style savings instruments
These work not because of high expected returns, but because humans overweight small probabilities of large gains.
Banks design these products to:
- encourage saving
- stabilize funding
- reduce withdrawal volatility
2.4 Cards, Payments, and Transaction Services
Payment products:
- debit cards
- credit cards
- transfers
- digital wallets
These services:
- generate fee income
- produce high-quality behavioral data
- anchor customers to the ecosystem
In modern banking, payments are data engines, not just utilities.
2.5 Insurance and Investment Products
Banks often distribute:
- insurance
- mutual funds
- structured investments
These products:
- diversify revenue away from interest income
- shift risk from the bank to customers
- require strong suitability and compliance controls
3️⃣ Interest Rates, Fees, and Pricing Power
Bank pricing reflects three forces:
- Cost of funding
- Risk of the borrower
- Regulatory constraints
Fixed rates exist to provide certainty. Floating rates exist to manage interest-rate risk.
Fees compensate for:
- operational cost
- capital usage
- non-credit risk
Pricing is never purely competitive—it is always risk-adjusted.
4️⃣ Risk: The Central Nervous System of Banking
Every banking decision is a risk decision.
Key risk dimensions include:
- Credit risk (default)
- Liquidity risk (cash shortage)
- Market risk (rates, FX)
- Operational risk (systems, fraud)
- Model risk (wrong assumptions)
Two core quantities dominate credit thinking:
- Probability of Default (PD)
- Loss Given Default (LGD)
Banks survive not by avoiding risk, but by measuring, pricing, and absorbing it.
5️⃣ Capital, Buffers, and Survival
Banks hold capital not to earn returns, but to survive rare, catastrophic losses.
Capital adequacy exists because:
- expected losses are manageable
- unexpected losses are existential
This is why banks:
- stress test
- plan for recessions
- grow slowly
A fast-growing bank is often a fragile bank.
6️⃣ Human Behavior in Finance
Purely rational models fail in banking.
Humans:
- dislike losses more than they like gains
- overborrow in good times
- panic in bad times
This is known as loss aversion.
As a result:
- defaults cluster
- liquidity dries up suddenly
- trust collapses faster than spreadsheets predict
Banking models must respect psychology, not just statistics.
7️⃣ AI and Data Inside Banks
AI adds value when it:
- improves risk estimation
- detects anomalies and fraud
- reduces operational friction
But banking data is:
- sensitive
- biased
- non-stationary
Accuracy alone is insufficient.
Models must be:
- explainable
- auditable
- robust under stress
This is why human-in-the-loop systems dominate financial AI.
8️⃣ Regulation and Responsibility
Banks are regulated not because they are inefficient, but because they are systemically important.
Failures propagate. Mistakes scale.
Regulation enforces:
- capital discipline
- consumer protection
- systemic stability
In banking, innovation must move with trust, not ahead of it.
9️⃣ Banking as Infrastructure
Banks are not just businesses. They are economic infrastructure.
They:
- allocate capital
- smooth consumption
- absorb shocks
- enable opportunity
This is why profitability is not the only metric. Stability, fairness, and resilience matter just as much.
🔚 Closing Thought
If you work in banking—especially with AI—you are not optimizing a model.
You are shaping:
- incentives
- behavior
- risk distribution
- trust
Banking is where mathematics meets society.
And society is always the harder system to model.
🧮 The Mathematics of Banking: Equations You Must Know
Modern banking is not powered by intuition. It is powered by equations.
If you work in banking—as an analyst, AI researcher, or system designer— these are the mathematical structures you are expected to think in.
This chapter is a map of the math behind real banking decisions.
1️⃣ Time Value of Money (TVM)
Everything in banking starts here.
Money today is worth more than money tomorrow.
$$ PV = \frac{FV}{(1 + r)^t} $$
Where:
- ( PV ) = present value
- ( FV ) = future value
- ( r ) = interest rate
- ( t ) = time
Loans, deposits, bonds, projects — all reduce to this idea.
2️⃣ Compound Interest
Banks do not earn linear returns.
$$ FV = PV (1 + r)^t $$
Compounding explains:
- why long-term loans are expensive
- why early repayments matter
- why maturity structure is critical
3️⃣ Loan Amortization (Fixed Payment Loans)
Mortgage math lives here.
$$ PMT = \frac{P r (1+r)^n}{(1+r)^n - 1} $$
Where:
- ( P ) = loan principal
- ( r ) = periodic interest rate
- ( n ) = number of payments
This equation defines:
- monthly installments
- interest vs principal split
- borrower affordability
4️⃣ Effective Interest Rate (EIR)
Banks care about true cost, not advertised rates.
$$ EIR = (1 + \frac{r}{m})^m - 1 $$
Used to:
- compare loan products
- standardize pricing
- comply with regulation
5️⃣ Expected Value (EV)
Credit decisions begin with expectations.
$$ \mathbb{E}[X] = \sum_i p_i x_i $$
But in banking, expected value is not enough.
Rare losses dominate survival.
6️⃣ Probability of Default (PD)
The backbone of credit risk.
$$ PD = P(\text{Default} \mid \text{Borrower Characteristics}) $$
Estimated using:
- logistic regression
- scorecards
- machine learning models
7️⃣ Loss Given Default (LGD)
Default is not total loss.
$$ LGD = 1 - \frac{\text{Recovery Value}}{\text{Exposure}} $$
Collateral, legal systems, and timing all matter here.
8️⃣ Exposure at Default (EAD)
How much is actually at risk.
$$ EAD = \text{Outstanding Balance at Default} $$
For credit cards and overdrafts, this is stochastic, not fixed.
9️⃣ Expected Credit Loss (ECL)
The core accounting equation of modern banks.
$$ \text{ECL} = PD \times LGD \times EAD $$
This single formula:
- drives provisioning
- impacts profit
- influences lending appetite
🔟 Risk-Adjusted Return on Capital (RAROC)
Banks do not maximize profit. They maximize risk-adjusted profit.
$$ RAROC = \frac{\text{Expected Return}}{\text{Economic Capital}} $$
If RAROC < hurdle rate → reject the loan.
1️⃣1️⃣ Capital Adequacy Ratio (CAR)
Survival constraint, not optimization.
$$ CAR = \frac{\text{Tier 1 + Tier 2 Capital}}{\text{Risk-Weighted Assets}} $$
Banks fail when capital buffers are insufficient, not when models are inaccurate.
1️⃣2️⃣ Risk-Weighted Assets (RWA)
Not all assets are equal.
$$ RWA = \sum_i w_i \cdot A_i $$
Where ( w_i ) reflects regulatory risk weights.
1️⃣3️⃣ Liquidity Coverage Ratio (LCR)
Can the bank survive a short-term shock?
$$ LCR = \frac{\text{High-Quality Liquid Assets}}{\text{Net Cash Outflows (30 days)}} $$
Liquidity crises kill banks faster than credit losses.
1️⃣4️⃣ Net Interest Margin (NIM)
Core profitability metric.
$$ NIM = \frac{\text{Interest Income} - \text{Interest Expense}}{\text{Earning Assets}} $$
Small changes in rates → massive balance-sheet effects.
1️⃣5️⃣ Duration (Interest Rate Risk)
Sensitivity to rate changes.
$$ D = \frac{\sum t \cdot PV(CF_t)}{PV} $$
Banks manage duration gaps to survive rate shocks.
1️⃣6️⃣ Stress Testing
Banking math under fear, not averages.
$$ \text{Loss}_{stress} = f(\text{GDP} \downarrow, \text{Unemployment} \uparrow) $$
Stress tests ask:
What if everything goes wrong at once?
1️⃣7️⃣ Expected Utility Theory
Rational decision-making baseline.
$$ \mathbb{E}[U(W)] = \sum_i p_i U(W_i) $$
But humans do not behave this way.
1️⃣8️⃣ Prospect Theory (Loss Aversion)
Real human behavior.
$$ U(x) = \begin{cases} x^\alpha & x \ge 0 \ -\lambda (-x)^\beta & x < 0 \end{cases} $$
Where lambda > 1 explains why losses hurt more than gains.
1️⃣9️⃣ Portfolio Diversification
Why banks lend to many borrowers.
$$ \sigma_p^2 = w^\top \Sigma w $$
Correlation kills diversification. Concentration kills banks.
2️⃣0️⃣ Model Risk (Reality Check)
All models are approximations.
$$ \text{Decision Risk} \neq \text{Model Error} $$
The biggest danger:
- wrong assumptions
- stable-looking data
- ignored tail events
🔚 Final Thought
If you understand these equations, you do not just understand banking math.
You understand:
- how banks survive
- why products are priced the way they are
- why AI must be cautious
- why trust matters more than accuracy
Banking is applied mathematics under uncertainty.
And uncertainty is never fully modelable.
🎯 50 Interview Q&A (Basic)
💡 Interview tip: Answers are written to sound systems-level, policy-aware, and AI-literate.
Each answer is hidden — open only when you want to rehearse.
Product & Banking Basics
1. What is the difference between deposits and funding?
Deposits are customer money entrusted to the bank, while funding is the broader set of liabilities the bank uses to finance assets, including deposits, bonds, and interbank borrowing. Deposits are typically cheaper and more stable.
2. Why do banks prefer long-term deposits?
Long-term deposits reduce liquidity risk and maturity mismatch, allowing banks to lend long-term without constant refinancing pressure.
3. What is NPA?
NPA (Non-Performing Asset) is a loan that stops generating income because the borrower fails to repay as scheduled, directly impacting capital and profitability.
4. Why is collateral important?
Collateral reduces loss given default (LGD) and aligns borrower incentives, transforming credit risk into asset-backed risk.
5. Why do banks care about liquidity?
Because even a solvent bank can fail if it cannot meet short-term obligations. Liquidity preserves trust.
6. Difference between retail and SME loans?
Retail loans rely on personal income and behavior data, while SME loans depend on cash flow, business cycles, and often collateral—making SMEs riskier but economically critical.
7. Why lottery savings work?
They exploit loss aversion and probability weighting: people prefer a small chance of a big reward over a guaranteed small return.
8. Why fixed rates exist?
They provide certainty to borrowers and are useful when interest-rate volatility is high.
9. Why floating rates exist?
They transfer interest-rate risk from banks to borrowers, protecting bank margins in changing macro conditions.
10. What is responsible lending?
Lending that balances access to credit with borrower affordability, long-term welfare, and systemic stability.
Risk & Economics
11. What is PD / LGD?
PD is the probability a borrower defaults; LGD is the percentage loss if default occurs. Together they define expected credit loss.
12. What is leverage?
Leverage is using borrowed funds to amplify returns—and losses. High leverage increases fragility.
13. Why diversification matters?
It reduces idiosyncratic risk by ensuring not all assets fail simultaneously.
14. What is capital adequacy?
A buffer that absorbs losses, ensuring banks remain solvent during stress.
15. Why banks survive crises?
Because of diversification, capital buffers, regulatory support, and central bank liquidity.
16. What is moral hazard?
When protection (e.g., bailouts) encourages riskier behavior because losses are partially externalized.
17. Why households default?
Income shocks, over-leverage, behavioral biases, and lack of financial buffers.
18. What is tail risk?
Low-probability, high-impact events that standard models often underestimate.
19. Why expected value is not enough?
Because humans and institutions care about downside risk, volatility, and survival—not just averages.
20. What is loss aversion?
The tendency for losses to hurt more than equivalent gains feel good, shaping financial behavior.
AI / Data Perspective
21. Where does AI add value in banking?
Risk assessment, fraud detection, personalization, operational efficiency, and policy evaluation.
22. What data is most sensitive?
Personally identifiable information, transaction histories, and credit data.
23. How do you avoid bias in credit models?
Through representative data, fairness constraints, explainability, and continuous monitoring.
24. What is model risk?
The risk that a model is wrong, misused, or applied outside its valid context.
25. Explain explainable AI to a banker.
Models must justify decisions in human terms so regulators, auditors, and customers can trust them.
26. Why accuracy is not enough?
Because false negatives, fairness, and stability matter more than raw prediction scores.
27. What is concept drift?
When the data-generating process changes, causing models trained on past data to degrade.
28. How do you test under recession?
By stress testing models with adverse macroeconomic scenarios and simulated shocks.
29. Why regulation matters for AI?
Because financial AI decisions affect livelihoods, stability, and trust.
30. What is human-in-the-loop?
A system where humans oversee, validate, and intervene in AI-driven decisions.
GSB / Policy Context
31. Why GSB is different from commercial banks?
GSB prioritizes social welfare and policy goals over profit maximization.
32. What is a policy bank?
A bank designed to implement government economic and social policies.
33. How does GSB reduce inequality?
By expanding access to savings and credit for underserved populations.
34. Why low-interest loans matter?
They prevent debt traps and support long-term economic resilience.
35. Why government-backed credit exists?
To address market failures where private banks avoid socially important risks.
36. How do policy goals affect risk?
They may increase credit risk but reduce societal risk.
37. Why profitability is not the only KPI?
Because public banks measure success through social impact and stability.
38. What is financial inclusion?
Ensuring everyone can access basic financial services.
39. What is informal debt?
Unregulated lending, often high-interest, outside the banking system.
40. Why digital banking matters?
It lowers costs, improves access, and generates data for better policy design.
Mindset
41. Why banking is a systems problem?
Because individual decisions aggregate into systemic outcomes.
42. Why math alone fails?
Because humans are not rational agents.
43. Why humans behave irrationally?
Due to cognitive biases, emotions, and limited information.
44. Why trust is central?
Banking collapses without confidence.
45. Why defaults are expected?
Because uncertainty is inherent in lending.
46. Why banks design contracts carefully?
Contracts encode incentives and risk-sharing.
47. Why transparency matters?
It sustains trust and accountability.
48. Why ethics matter?
Because financial decisions affect lives at scale.
49. Why banking changes slowly?
Because stability is more valuable than speed.
50. Why this job matters?
Because banking shapes economic opportunity and social resilience.
🎯 50 Interview Q&A (Banking & Finance)
💡 How to use this blog:
Read the questions first, think like a banker / economist / data scientist,
then open the answers to calibrate your thinking, not memorize sentences.
This is about judgment, risk awareness, and systems thinking.
Product & Banking Basics
1. What is the difference between deposits and funding?
Deposits are customer money placed with the bank, usually stable and low-cost.
Funding refers to all sources of money a bank uses, including deposits, bonds, and interbank borrowing.
2. Why do banks prefer long-term deposits?
They reduce liquidity risk and maturity mismatch, allowing banks to lend long-term safely.
3. What is a non-performing asset (NPA)?
A loan that no longer generates income because the borrower has failed to meet repayment obligations.
4. Why is collateral important?
Collateral reduces losses when borrowers default and aligns incentives between borrower and lender.
5. Why do banks care so much about liquidity?
Because a bank can be solvent but still fail if it cannot meet short-term obligations.
6. Difference between retail and SME loans?
Retail loans depend on individual income and behavior; SME loans depend on business cash flows and economic cycles.
7. Why do lottery-based savings products work?
They leverage behavioral biases: people overweight small probabilities of large rewards.
8. Why do fixed interest rates exist?
They provide payment certainty to borrowers and are attractive during volatile interest-rate environments.
9. Why do floating interest rates exist?
They allow lenders to pass interest-rate risk to borrowers and protect margins.
10. What is responsible lending?
Granting credit based on affordability, long-term borrower welfare, and systemic stability.
Risk & Economics
11. What are PD and LGD?
Probability of Default (PD) measures how likely a borrower will default;
Loss Given Default (LGD) measures how much is lost if default occurs.
12. What is leverage?
Using borrowed money to amplify returns, which also amplifies losses.
13. Why does diversification matter?
It reduces exposure to individual failures by spreading risk across assets.
14. What is capital adequacy?
The level of capital a bank holds to absorb unexpected losses and remain solvent.
15. Why do banks usually survive financial crises?
Because of capital buffers, diversification, regulation, and central bank support.
16. What is moral hazard?
When protection from losses encourages riskier behavior.
17. Why do households default on loans?
Income shocks, excessive leverage, poor financial planning, and behavioral biases.
18. What is tail risk?
Rare events with extreme consequences that standard models often underestimate.
19. Why is expected value not enough in finance?
Because people and institutions care about downside risk and survival, not just averages.
20. What is loss aversion?
Losses psychologically hurt more than equivalent gains feel good.
AI & Data Perspective
21. Where does AI add value in banking?
Credit scoring, fraud detection, customer segmentation, and operational efficiency.
22. What data is most sensitive in banking?
Personal identity data, transaction records, and credit histories.
23. How do you reduce bias in credit models?
By using representative data, fairness constraints, explainability, and monitoring.
24. What is model risk?
The risk that a model is incorrect, misapplied, or used outside its valid context.
25. Explain explainable AI to a banker.
Models must justify decisions clearly so humans can trust, audit, and regulate them.
26. Why is accuracy not enough?
Because false negatives, fairness, robustness, and stability matter more.
27. What is concept drift?
When relationships in the data change over time, degrading model performance.
28. How do you test models under recession?
By stress testing with adverse economic scenarios and simulated shocks.
29. Why does AI in finance require regulation?
Because automated decisions affect livelihoods and systemic stability.
30. What is human-in-the-loop?
A system where humans oversee, validate, and intervene in AI decisions.
Policy & System-Level Thinking
31. Why do governments intervene in credit markets?
To correct market failures and support economic stability.
32. What is a policy-driven financial institution?
An institution designed to serve economic or social objectives beyond profit.
33. How does finance affect inequality?
Access to credit can amplify opportunity or deepen inequality.
34. Why do low-interest loans matter?
They reduce debt burdens and improve long-term repayment sustainability.
35. Why does government-backed credit exist?
To enable lending where private risk appetite is insufficient.
36. How do policy goals change risk management?
They trade higher financial risk for lower societal risk.
37. Why is profitability not the only metric?
Because financial systems exist to support real economic activity.
38. What is financial inclusion?
Ensuring everyone can access basic financial services.
39. What is informal debt?
Unregulated lending outside the formal banking system.
40. Why does digital banking matter?
It lowers costs, expands access, and improves data-driven decision-making.
Mindset & Philosophy
41. Why is banking a systems problem?
Individual decisions aggregate into systemic outcomes.
42. Why does math alone fail in finance?
Because humans are not perfectly rational.
43. Why do people behave irrationally with money?
Due to biases, emotions, and limited information.
44. Why is trust central to banking?
Without trust, deposits and credit collapse.
45. Why are defaults expected?
Because uncertainty is inherent in lending.
46. Why are contracts carefully designed?
They encode incentives and risk-sharing.
47. Why does transparency matter?
It supports trust, accountability, and stability.
48. Why do ethics matter in finance?
Because financial decisions affect lives at scale.
49. Why does banking evolve slowly?
Because stability is more valuable than speed.
50. Why does working in banking matter?
Because finance shapes economic opportunity and resilience.