Good Debt vs. Bad Debt: The Difference Nobody Explains
Credit cards that charge interest average 22.4% APR. A 30-year mortgage runs about 7.4%. Same word, different math. Here's the 4-question test for telling good debt from bad.

Good debt is borrowing money to buy something that earns more value than the loan costs in interest. Bad debt is borrowing to buy something that loses value while you pay interest on it. The Federal Reserve Bank of New York puts the return on a typical bachelor's degree at 12.5% a year, well above the roughly 8% long-run return of the stock market. Meanwhile, credit cards that charge interest average 22.36% APR and a typical payday loan carries an APR of nearly 400%. Same word, different math.
Total U.S. household debt stood at $18.8 trillion in the second quarter of 2026, per the New York Fed. Mortgages account for $13.12 trillion of it. Credit cards account for $1.26 trillion. Both got called "debt" in the headline. They are not the same thing.
Every personal-finance article you've ever read gives you the same answer. Mortgage good. Credit card bad. Then it stops, like that's enough. It isn't. The truth is messier, and the rule of thumb fails in most real-world borrowing decisions you'll actually face.
Good debt's defining trait is a positive yield gap: the asset returns more than the loan costs. Bad debt's defining trait is a negative yield gap that compounds against you. A typical bachelor's degree returns about 12.5% a year on its cost (NY Fed, 2025), well above the roughly 8% long-run return of stocks. Credit cards that charge interest average 22.36% APR (Federal Reserve G.19, August 2026) and a typical payday loan's APR is nearly 400% (CFPB). Same word, completely different math.
This article gives you the 4-question test that tells you whether a debt is working for you or against you, the math behind why most "good" debts can flip to bad, and the cases where bad debt is the only sane move.
Read more: The Debt Trap: What Nobody Told You About Borrowing Money | The Minimum Payment Trap
What Is Good Debt?
Good debt is borrowing money to buy something that earns more value (in price, income, or future earnings) than the loan costs in interest. The classic examples are mortgages, federal student loans for degrees with positive labor-market returns, and small-business loans on a venture that out-earns the cost of capital. The defining trait isn't the category. It's the math.
A 2025 analysis from the Federal Reserve Bank of New York found the return on a bachelor's degree was 12.5% in 2024, which it called "easily exceeding the threshold for a sound investment." For comparison, the New York Fed notes the stock market has returned about 8% over the long run. Most degrees clear that bar. Some clear it by a wide margin. Others don't clear it at all.
Mortgages are the other classic example. U.S. home prices have risen roughly 4% a year since 1987, per the Case-Shiller national index. U.S. mortgaged borrowers collectively held about $17.9 trillion in home equity in the second quarter of 2026, with the average mortgaged household sitting on roughly $310,000, per Cotality. Add the rent you don't pay (because you live in the house) to the price appreciation, and a mortgage at today's 7.4% on a primary home held for 10+ years can still clear the math.

Small-business loans work the same way when they work at all. Borrow at 8%, run a business that returns 20% on equity, the spread is yours. The bank is willing to lend at that rate because their underwriting says you're likely to clear it. That's information worth listening to. When a regulated lender will give you 30 years to pay something back at a fixed rate, the system is signaling that they expect both you and the asset to still be standing at the finish line.
That's the test most "good debt" headlines skip. The label isn't the answer. The expected return on the underlying asset is the answer. A degree program with a placement rate near zero isn't good debt regardless of what the loan officer calls it. A house in a neighborhood that's losing population isn't good debt regardless of what Zillow says it's worth today. The category sets the ceiling. The conditions decide whether you actually hit it.
Notice what working examples of good debt have in common. The asset earns or grows. The loan ends. The thing you bought is still worth something when the last payment clears. The math has a positive sign on it the whole way through. That's the picture of good debt working as advertised. Most articles stop here. We're not going to.
12.5%
Annual return on a typical U.S. bachelor's degree in 2024, well above the stock market's roughly 8% long-run return
Federal Reserve Bank of New York, 2025
What Is Bad Debt?
Bad debt is borrowing money to buy something that loses value, generates no income, or charges you more in interest than the asset can ever return. Credit card balances, payday loans, BNPL on consumer goods, and most car loans on depreciating vehicles all qualify. The defining trait is that you keep paying long after the thing you bought has lost its value.
The numbers tell the story. The average APR on credit card accounts charged interest was 22.36% in August 2026, per the Federal Reserve's G.19 release. 47% of cardholders carry a balance month to month, per the 2026 Bankrate Credit Card Debt Report. Pay only the minimum on a $5,000 balance at that rate (1% of the balance plus interest, $25 floor) and you'll spend about $8,200 in interest and over 19 years to clear it.
Payday loans are worse. A typical two-week payday loan with a $15-per-$100 fee equates to an APR of nearly 400%, per the CFPB. The same agency found that payday borrowers are indebted a median of 199 days (55%) of the year. That's not borrowing. That's renting your own paycheck for over half the calendar year.
What about Buy Now, Pay Later? 47% of users made at least one late payment in the past year, and 63% have held multiple BNPL loans at the same time, per a 2026 LendingTree survey. The label says "Pay in 4." The financial reality says debt with late fees, and since fall 2025 FICO has offered lenders credit scores that can count BNPL loans.
The shared trait across credit cards, payday, and BNPL: the asset returns nothing. You can't sell groceries back. You can't refinance dinner. The interest is pure cost on something that already disappeared. That's the structural difference. With good debt, the asset is doing some of the work. With bad debt, the asset clocks out the moment you swipe the card. You're the only one still paying.
The chart below is the whole article in one picture. Each row is a common debt type. The red bar is what the loan costs you. The green bar (when it exists) is what the asset returns. Where the gap goes the wrong way, you're going backward. Where it doesn't, you're going forward.
The Debt Yield Gap
Good debt buys something that grows faster than the loan costs. Bad debt pays interest on something that shrinks or is already gone.
Sources: Freddie Mac (Oct 8, 2026); Federal Student Aid 2026-27; New York Fed, Liberty Street Economics (Apr 2025); Edmunds Q3 2026; Kelley Blue Book via Experian (car values); Federal Reserve G.19 (Aug 2026); CFPB (payday). Returns are typical yearly figures, not guarantees.
What's the 4-Question Debt Test?
Run any borrowing decision through these four questions before signing. If the debt fails one, it's borderline. If it fails two, it's bad debt with extra steps.
1. Will the asset's value or earnings grow faster than the after-tax interest rate? This is the yield gap. A 7.4% mortgage on a home appreciating about 4% per year, plus the rent you don't pay, can clear it. A 22.4% credit card balance on groceries doesn't.
2. Can you pay the loan off before the asset is gone, used up, or worthless? A 30-year mortgage on a house you'll live in for 15+ years passes. An 84-month auto loan on a car that's worth half its purchase price by year four fails.
3. Would you still buy this if the price was the loan total in cash, today, with no financing? This is the "cash test." It strips out the payment-plan psychology. If the answer is no, the financing isn't helping you. It's hiding a no from yourself.
4. Does the monthly payment crowd out savings or investment that would compound at a higher rate? A $400 car payment that prevents you from investing $400 a month in a long-term portfolio is a hidden negative return. Don't only count what the loan costs. Count what you couldn't do with the money instead.
Run a real number through it. A $30,000 used car at 10.6% APR (Edmunds' Q3 2026 used-car average) over 72 months totals roughly $40,700 paid. The car is worth around $10,000 by payoff. Yield gap: deeply negative. Test fails 1, 2, and 4. Verdict: bad debt with a steering wheel.
Run a counter-example. A $200,000 mortgage at 7.4% on a primary home you'll live in for 10+ years. The asset appreciates about 4% a year, plus you skip rent. Yield gap is negative on price alone, positive once you count the rent saved. Test passes 1 and 2, conditional on the housing payment staying below 28% of gross income. Verdict: good debt, conditionally.
Here are those two loans side by side with the two other debts most people carry, scored the same way:
| Debt | 1. Asset outgrows the rate? | 2. Paid off before it's gone? | 3. Cash test | 4. Room left to invest? | Verdict |
|---|---|---|---|---|---|
| $30,000 used car, 10.6%, 72 months | No: about $40,700 paid for a car worth about $10,000 | No | Your call | No | Bad debt |
| $200,000 mortgage, 7.4%, home kept 10+ years | Yes, once the rent you skip is counted | Yes | Your call | Yes, if housing stays under 28% of gross income | Good, conditionally |
| Credit card balance at 22.4% on groceries | No: the groceries return nothing | No | Your call | No: the interest crowds out saving | Bad debt |
| Federal loan for a finished bachelor's degree | Usually: about 12.5% a year on average (NY Fed) | Yes, the degree does not wear out | Your call | Depends on the payment | Good, if you finish |
The cash test is the one only you can score, because it depends on what the thing is worth to you. The other three are arithmetic.
Most personal-finance content treats every mortgage as good debt and every credit card as bad debt. The 4-question test reveals the truth: a mortgage you can't afford fails the same way a credit card balance does, just on a longer timeline. And a credit card paid in full every month is a tool, not a trap.
When Does Good Debt Become Bad Debt?
Any "good debt" can flip categories under specific conditions. Mortgages flip when the housing payment exceeds 28% of gross income, or when the home doesn't appreciate, or when the buyer holds it for less than 5 years and gets eaten by closing costs and realtor fees. Student loans flip when the borrower never finishes the degree, or the program's labor-market value doesn't justify the borrowed amount.
Auto loans flip the moment the loan term outlasts the car's useful life, or the buyer rolls negative equity from the last car into the new one. And that's not theoretical. It's what's actually happening at scale right now.
30.9% of trade-ins toward new vehicles in Q1 2026 carried negative equity, the second-highest quarter on record and the highest since early 2021, per Edmunds' Q1 2026 Insights Report. The average underwater borrower owed $7,183 on a car they no longer wanted, the highest ever for a first quarter. 26% of underwater trade-ins carried $10,000 or more in negative equity.

Here's the math that should get your attention. The average new-car payment for a buyer rolling negative equity into the loan is $932 a month, $159 more than a typical buyer, per Edmunds. 43% of those negative-equity new-car purchases are financed with 84-month loans. That's seven years of payments on a vehicle that loses roughly 35% to 40% of its value in the first three years and more than half by year five, per Kelley Blue Book figures cited by Experian. By month 60, you're still two years from done on a car worth less than half what you paid.
Now stretch the math out. A $40,000 new car at 7.0% (Edmunds' Q3 2026 new-car average) over 84 months totals roughly $50,700 paid. The car is worth around $14,000 by payoff. So you'll send $50,700 to a lender and end up holding $14,000 of asset. That's a $36,700 net loss on the trade, before fuel, insurance, repairs, or registration. Run that through the 4-question test and the only question it passes is "did the dealer have your signature on the line." It fails 1, 2, and 4 by a landslide.
The same loan structure on a 36-month term, on a car you actually keep for 10 years, looks different. The interest cost is less than half, about $4,500 instead of $10,700. The depreciation curve flattens by year four. By year seven you're driving a fully owned car, the payment has been redirected into something useful, and the math swings to neutral or positive. Same APR. Same dealership. Different conditions, different verdict.
What started as a "good debt" auto loan with a single trade-in flipped to bad debt. Then to worse debt. Then to a trap with a steering wheel and a payment plan that outlives the car. Same product. Different math depending on the conditions.
30.9%
Share of Q1 2026 U.S. trade-ins toward new vehicles carrying negative equity, the highest since early 2021
Edmunds Q1 2026 Insights Report
The rule isn't "auto loans are bad." The rule is: auto loans become bad the moment the loan outlasts the car, the term exceeds 60 months, or you ever consider rolling negative equity forward.
Why "Bad" Debt Is Sometimes the Only Move
Bad debt becomes survival debt when there's no alternative. A medical emergency. A tire to keep getting to work. A utility shutoff with kids in the house. Calling that "bad debt" is technically correct and practically useless. The Untaught take: name what you're doing, plan to retire it as fast as possible, and don't repeat the borrow.
This is where the emergency-fund gap shows up. Only 46% of Americans have a 3-month emergency fund, down from 53% in 2021, per the FINRA Foundation's National Financial Capability Study (2024 wave, released July 2025). Without that buffer, every surprise becomes a high-APR borrow, which is why living paycheck to paycheck and bad debt so often show up together. The bad-debt cycle doesn't start with bad decisions. It starts with no alternative.
What to do if you're already there? Stop the bleed first. Credit card minimum payments are not a plan. They are the trap setting itself. Build a $500 to $1,000 starter emergency fund before you do anything else, even before aggressive payoff, so the next surprise doesn't restart the cycle. Then attack the highest-APR balance with everything you've got.
This is the same arithmetic that makes the minimum payment trap so effective. It's not about discipline. It's about cash flow. Without a starter cushion, you can't escape the cycle, no matter how aggressively you pay.
How Do Most People Mix Good and Bad Debt?
The average American household carries both. $13.12 trillion in mortgages (good, on average). $1.26 trillion in credit cards (bad, on average). $1.65 trillion in student loans (mixed, depending on the degree). $1.71 trillion in auto loans (mixed, increasingly bad as terms lengthen). Treating all of it as one big pile is how the system wins.
Where $18.8 Trillion in Household Debt Lives
U.S. household debt by type, Q2 2026. Mortgages are 70%. Credit cards are under 7% but carry by far the highest rates.
Source: Federal Reserve Bank of New York, Quarterly Report on Household Debt and Credit, Q2 2026 (released August 11, 2026).
About 70% of total U.S. household debt is mortgages, per the New York Fed Q2 2026 Household Debt and Credit Report. That share has been roughly stable for years. Over the year to Q2 2026, credit card balances grew $54 billion (about 4.5%) and auto loans $58 billion (about 3.5%), both far faster than mortgages (about 1.4%). The composition isn't shifting to good debt. It's shifting to bad debt and longer-term car debt, in the same households.
So the question for your own balance sheet isn't "do I have debt." Almost everyone does. The question is "what's the mix, and is it improving or getting worse?" If your bad-debt total is climbing year over year while your good-debt total is steady, you're in the wrong direction. If you're paying down credit cards while your mortgage shrinks on schedule, the trajectory is right.
Prioritize the bad-debt payoff first by APR (highest rate first), then by balance. The math return on knocking out a 22.4% credit card is the same as a guaranteed 22.4% return on an investment. Nothing in the public market beats that consistently. Pay off the high-APR debt, and the freed-up payment becomes the seed of a real DCA habit.
That's the redirect Untaught keeps coming back to. The system designed credit cards, payday loans, and BNPL to extract a recurring payment from you every month. When that payment is gone, the muscle is already trained. Send the same dollar amount somewhere that compounds for you instead.
How Do You Use This Right Now?
List every debt you have. Write the balance, APR, and what you originally bought. Run each one through the 4-question test. Anything that fails is your highest priority. Set a number on it: the date you intend it to be zero.
Then redirect the cleared payment into a long-term DCA habit. The same $300 a month that was killing you in credit card interest will work for you when compound interest is going the right way. The S&P 500 has averaged about 10% per year since 1928, per NYU Stern. A $300 monthly DCA at an 8% annual return grows to roughly $54,000 in 10 years (monthly deposits, FV = C × ((1 + r)^n − 1) ÷ ((1 + r)^(1/12) − 1), the model used across Untaught). That's the same payment, on the other side of the trade.
Here is what a few common cleared payments become on that same model, monthly deposits growing 8% a year, compounded monthly at the equivalent rate:
| Payment you free up | You put in over 10 years | Worth after 10 years | You put in over 20 years | Worth after 20 years |
|---|---|---|---|---|
| $100 a month | $12,000 | $18,000 | $24,000 | $56,900 |
| $300 a month (a credit card payment) | $36,000 | $54,000 | $72,000 | $170,700 |
| $400 a month (a car payment) | $48,000 | $72,000 | $96,000 | $227,600 |
| $932 a month (Edmunds' average negative-equity new-car payment) | $111,840 | $167,900 | $223,680 | $530,300 |
The same $400 monthly payment that's draining you on a 7-year auto loan can grow to roughly $72,000 in 10 years if you redirect it into a DCA habit at an 8% average return. That's not a trick. That's the math the system is designed to keep you from running.
You can run your own numbers in the DCA Calculator. Plug in the payment you'd free up. Pick a time horizon. See what it becomes. The output is the trade you're actually making every month you keep the bad debt.
Don't miss the rest of the story.
Get a heads up when the Untaught app launches. Free, and no spam.
Frequently Asked Questions
Compare the after-tax interest rate to what the asset will earn or grow over the loan period. If the asset grows faster, the debt is working for you. If the loan costs more than the asset returns, you're paying for the privilege of going backward. Stocks have returned about 8% over the long run by the New York Fed's yardstick, and a 22.4% credit card balance will never be beaten by that.
It depends on the degree, completion, and labor-market return. The Federal Reserve Bank of New York's 2025 analysis puts the return on a typical bachelor's degree at about 12.5% a year. Programs the borrower never finishes, or degrees with low labor-market value, can flip a federal loan into bad debt fast. For more, see our guide on what student loans actually cost over 10 years.
No. A mortgage is good debt only when the payment fits within sustainable household budget limits (commonly 28% of gross income for housing) and the home is held long enough to outlast transaction costs. Mortgages on overstretched payments, short holding periods (under 5 years), or in stagnating local markets can fail the 4-question test as badly as a credit card balance.
Most are. Cars depreciate roughly 15-20% in year one. The average new-car APR was 7.0% and used 10.6% in Q3 2026, per Edmunds. When loan terms exceed 60 months, or borrowers roll negative equity into the next loan, the math gets ugly fast. 30.9% of Q1 2026 new-car trade-ins were already underwater (Edmunds), the highest share since early 2021. A 60-month or shorter loan on a car you'll keep 10 years can pass the test. Most actual auto loans don't.
Almost never as a balance. A credit card used as a payment method and paid in full every month is a tool. A credit card carrying a 22.4% APR balance is bad debt, full stop, and the math doesn't change because the rewards card has a logo on it. 47% of cardholders carry a balance, per Bankrate's 2026 Credit Card Debt Report. That's not a personal finance failure. That's the companies who profit from financial ignorance doing exactly what the model is built to do.
The "mortgage = good, credit card = bad" rule of thumb is half an answer. The Federal Reserve, the New York Fed, the CFPB, and Edmunds all publish data that says the same thing: every debt category contains both kinds. The label on the loan isn't the answer. The math is.
Run your debts through the 4-question test. Pay off whatever fails. Redirect the cleared payment into something that compounds for you instead of against you. That's not a trick. That's the part the system never wanted to teach you.
Start here: This article is part of The Debt Trap, our complete guide to how the system keeps you borrowing and what to do about it.
Next step: If you're already carrying high-APR debt, read how to get out of debt when you're living paycheck to paycheck for the practical playbook. When you're ready to redirect, see what $20 a week becomes over 10 years, or run your own numbers in the DCA Calculator.
This article is for educational purposes only and does not constitute financial advice. Untaught does not hold, move, or custody any funds. Past performance does not guarantee future results. Always do your own research before making investment decisions.
Quick calculator
Your coffee money could have become
$19,233
from $9,900 invested
Monthly buys, Oct 2021 to Sep 2026, at Coin Metrics month-end closes. Past results, not a forecast.