how risk models are used to make consumer credit lending decisions, and how they make money

Welcome! I am glad you are here. To help me tailor this explanation, could you tell me a bit about your background? Are you looking to understand this from a consumer's perspective, a career interest, or perhaps an academic one?

I have a career interest in fintech or banking.

It is great to have you! With your sights set on fintech and banking, you are looking at one of the most data-driven, fast-paced corners of finance.

What is Credit Risk?

At its simplest, credit risk is the possibility that a borrower will fail to pay back their loan. Every time a bank or a fintech startup hands out money, they are taking a leap of faith. If the borrower defaults, the lender does not just lose out on the interest they hoped to make — they lose the actual cash they lent out, known as the . Because credit returns have a unique shape — your upside is capped at the interest rate you charge, but your downside is a 100% loss of your cash — keeping this risk under wraps is the single most important factor in whether a lender survives or goes bankrupt.

Just so you know, I'm a data scientist working for a fintech. I've got several years in financial services, but I've never worked on lending / credit risk side of the business. I have a good understanding of modelling techniques used. I'm more interested in how these models are combined / used to make money.

That is incredibly helpful context. Since you already understand the statistical modeling side, we can skip the basics of model training and focus on what you actually care about: how those models function as the engine of a highly profitable business.

At a fintech lender, models are not just predicting defaults in a vacuum. They are directly driving the levers that determine unit economics. To see this in action, we have to look at the absolute core formula of how a lender makes money.

The Unit Economics of a Loan

Every loan is a mini-business with its own revenue and expenses. To understand how models make money, we first have to understand the basic profit formula for a single loan:

$\text{Expected Profit} = \text{Expected Revenue} - \text{Expected Costs}$

\text{Expected Profit} = [I \times (1 - P_d)] + [F_r \times P_d] - [C_f + O_c + (P_d \times LGD)]

This is where your world as a data scientist meets the balance sheet. A credit risk model's main job is to estimate that key variable:

P_d

(Probability of Default). If your model is highly accurate, the business can do two powerful things to maximize profit. First, they can adjust the approval threshold to decline borrowers whose expected profit is negative. Second, they can use risk-based pricing — charging a higher interest rate (

I

) to higher-risk borrowers to compensate for the higher probability of default (

P_d

). If the model's prediction of

P_d

is off by even a few percentage points, the pricing strategy fails, and the portfolio can quickly become unprofitable.

Continue to how lenders make money on interest spread and fees

Let's crack open the engine of lender profitability. To understand how lenders make money on the spread, we have to look at their single most important profitability metric: the (NIM).

The Spread: Yield vs Cost of Funds

At its core, lending is a retail business, but instead of buying wholesale goods, you are buying wholesale money.

Your Cost of Funds ( $C_f$ ) is what you pay to get capital—whether that's paying $1.5\%$ interest to retail depositors or paying $6\%$ to an institutional investment fund that backs your fintech.

Your Yield is the average interest rate ( $I$ ) you charge your borrowers. The raw difference between the two is your interest spread. If you borrow money at $5\%$ and lend it out at $12\%$ to a prime borrower, you have a $7\%$ spread to cover your operational costs and credit losses.

Boosting Revenue: Non-Interest Fees

While interest is the main event, lenders rely heavily on non-interest revenue streams to supercharge their unit economics.

Origination Fees: An upfront fee (often $1\%$ to $8\%$ of the loan amount) deducted directly from the payout. If a customer takes out a $10,000 loan with a $5\%$ origination fee, they receive $9,500, but they still owe $10,000 plus interest. This is pure immediate revenue for the lender.
Late Fees: Charges levied when a borrower misses a payment window. While these are designed to incentivize good behavior, they also act as high-margin revenue.
Prepayment Penalties: Some lenders charge a fee if a borrower pays off their loan early. Why? Because early repayment cuts short the stream of interest income the lender had modeled into their profit projections.