Moving day. The boxes are in, the keys work, and the electricity company wants a deposit before the account goes live.

The amount is based on a credit check. The credit check shows no prior utility account, because this is the first apartment. No prior utility account is treated the same as a bad credit history. The deposit is often $100 to $300, sized to roughly two months of estimated billing. It is due upfront, usually within a few days of applying.

The deposit sits at the utility company for up to twelve months. If every bill is paid on time for a full year, it comes back, sometimes with interest, often without. In many states the rules around returning deposits are set by the public utility commission and vary by provider. In some states, enrolling in a qualifying low-income program waives the deposit requirement. Most renters opening a new account do not know this.

That deposit is gone on day one, before a single kilowatt-hour has been used.

The building the landlord never fixed

Renters who pay their own electric bills are roughly 87% of American renters. They also pay for a problem they did not create: a building the landlord had no financial reason to improve.

The issue has a name. Economists call it the split incentive. The landlord pays for upgrades, insulation, windows, and weatherstripping. The tenant pays the electric bill. That means the landlord saves nothing by improving the building, so the building does not get improved. Studies have found that renters use roughly 9% more electricity than non-renters in comparable locations once other differences are controlled for. That gap represents real money on the bill every month.

Low-income renters live disproportionately in older, less efficient housing. The result shows up in the numbers. According to the American Council for an Energy-Efficient Economy, in a September 2024 study, the median low-income household in the United States spends 8.3% of annual income on energy bills, more than the 6% threshold typically used to define a high energy burden. One in four low-income households spends more than 15% of income on energy.

Data from the Federal Reserve Bank of St. Louis, drawing on the 2024 Consumer Expenditure Survey, shows the gap in full. Low-income households spend 13.3% of pretax income on utilities. High-income households spend 2.5%. For electricity alone, the figures are 5.4% and 0.8%.

Same grid. Different cost as a share of income.

When a bill is too high

The bill comes on a fixed schedule. The paycheck does not always arrive at the same time. For a household spending 15% or more of its income on energy, a high bill in a cold month becomes a choice between the electric bill and something else.

Among households earning under $10,000 a year, roughly 32% received a disconnection notice at some point, according to analysis from the Bipartisan Policy Center. During a twelve-month period from May 2020 to May 2021, an estimated 1.5 million low-income households had their electricity cut off, according to research from Indiana University’s Energy Justice Lab.

Reconnection fees vary by state and provider. Where they exist, they fall on the households that can least afford them. One documented example from Detroit placed the fee at $185. What makes this worth noting is what reconnection fees cannot be paid with: federal LIHEAP energy assistance money covers bills and deposits, but it does not cover the cost of getting the lights turned back on after a shutoff.

The part nobody mentions

The Low Income Home Energy Assistance Program, known as LIHEAP, is the main federal safety net for utility bills. In fiscal year 2024, it served approximately 5.9 million families. The National Low Income Housing Coalition estimates this was about 17% of eligible households. More than eight out of ten eligible households received nothing.

LIHEAP funding was cut from $6.1 billion in fiscal year 2023 to $4.1 billion in fiscal year 2024. The National Energy Assistance Directors Association estimated that the reduction would leave approximately 1.5 million fewer households served than the year before. Local program waitlists opened and filled in hours.

The deposit, the drafty building, and the disconnection fee are small individually. Together, over the course of a year, they add up. The safety net designed to help covers fewer than one in five of the households it was built for.

What helps

A few things are worth knowing.

Enrolling in LIHEAP, SNAP, or Medicaid will waive the utility deposit requirement in a number of states. The utility company is not always required to tell applicants this. Calling the provider directly and asking, or checking the state public utility commission’s website, can confirm whether the waiver applies.

A letter of credit from a prior utility, showing twelve consecutive months of on-time payments at a previous address, is accepted in place of a cash deposit by many providers. Renters with a good payment history at an earlier address can call and ask for one.

If a deposit is not returned after twelve months of on-time payments, the state public utility commission is the place to file a complaint. The process is free, and in many states the commission is required to respond.

What it comes down to

Being a renter with a low income does not just mean paying rent and a utility bill. It means paying a deposit before the lights come on, paying higher bills for a building the landlord had no financial reason to improve, and running into fees and a strained safety net when a bill gets too large to cover.

If you have ever found yourself choosing between the electric bill and something else, you were not being careless. You were paying the premium the system charges people who do not have enough of a cushion to avoid it.

A notice arrives in the mail on a Thursday. It is a speeding ticket, 12 miles over the limit on a surface street. The base fine is $35. The total amount due, after all the state surcharges, court fees, and penalty assessments have been added, is $234. The due date is 30 days away.

For someone with $234 in a checking account and no bills coming, the choice is easy: pay it and move on. For someone with $234 and a utility bill and a co-payment due in the same month, the ticket goes to the bottom of the pile.

That decision, made on a Thursday evening over a stack of paper, is where $234 becomes something much larger.

What is in the total

The number on the face of a traffic ticket is not the amount the court will collect. In California, a speeding violation with a base fine of $35 carries a stack of additions: a state penalty assessment, a county assessment, a court construction fee, a DNA identification fund fee, an emergency medical services penalty. By the time those are added, the $35 base fine reaches approximately $234. The same mechanism applies in other states. The penalty assessment structure differs, but the principle holds: the fine you owe is not the fine on the notice.

This matters because the ticket is priced against the base fine in many people’s minds. Someone who hears “it is a $35 ticket” may not realize the actual bill is six times that. When they do, and when they cannot pay it, the clock on the next layer of costs starts.

What happens when the deadline passes

Courts treat an unpaid ticket as a separate problem from the original violation. In California, if a fine is not paid by the due date and no extension has been arranged, the court can add a civil assessment of up to $300. Some courts can also refer the debt to a collections agency, which adds its own fees on top. In Texas, an unpaid ticket can trigger a default judgment, a license suspension warning, and referral to a collections agency that takes a percentage of whatever is recovered.

The 30-day due date is not a soft deadline. It is the last point at which the original amount still applies.

A 2025 Urban Institute survey of working-age adults found that about one in five had their household charged with a fine or fee in 2024, and that traffic and parking tickets were the most common type. Among those charged, one in four reported still owing unpaid fines from 2024 or earlier. The survey also found that few of them had been asked about their ability to pay, or offered a reduced fine, a payment plan, or community service as an alternative.

The license

In more than half of states, an unpaid traffic fine can lead to a suspended driver’s license. The suspension is a debt collection mechanism. The driving violation and the unpaid debt are treated as a single problem by the same court, and the license is what gets taken until the debt is paid.

As of 2019, the Fines and Fees Justice Center, a legal research and advocacy organization, estimated that roughly 11 million driver’s licenses in the United States were suspended for debt-related reasons. That figure predates a wave of state reforms, and the current number is lower. But as of 2025, more than half of states still allow some version of debt-based suspension.

Reinstating a license after suspension for nonpayment typically requires paying the original fine, plus the civil assessment, plus a reinstatement fee that varies by state. The debt does not pause while the license is suspended. It waits.

The part nobody mentions

The assumption behind debt-based suspension is that it creates pressure to pay. The reality is that for people without the money to pay the original fine, losing a license makes it harder, not easier, to earn the money.

A 2026 study published in the Journal of Applied Psychology followed more than 800 Alabama residents and found that license suspension was associated with higher rates of unemployment. It was also linked to a shift toward informal, unregistered employment, work that does not appear on a resume or qualify someone for a loan. A separate analysis of 24 state reform efforts found that in places where debt-based suspension was removed, employment rates rose, with the clearest gains among non-white workers in states with comprehensive reform.

This is the loop: the fine a person cannot pay leads to a suspension that cuts off transportation. Without transportation in most of the United States, work that pays enough to clear the debt becomes harder to reach. The original $234, unpaid on a Thursday, compounds into a problem that now touches the job, not just the bank account.

What helps

The options exist, but they require asking before the deadline, not after.

Most courts allow a payment plan or an installment agreement, and in California some courts have formal ability-to-pay hearings where fines can be reduced or waived based on income. Community service is another option: some courts allow drivers to work off a portion of the fine. Texas law allows judges to waive or reduce fines when neither a payment plan nor community service is workable, though the standard for this is high.

The key detail is timing. These options, hearings, plans, service agreements, are most available before the due date passes. Once a civil assessment has been added and the debt has been sent to collections, the original fine is no longer the amount in dispute.

California courts have online tools for requesting an ability-to-pay review. Texas Law Help, a legal aid website, has a plain-language guide covering the options for someone who cannot afford a ticket in Texas. Similar resources exist in most states through legal aid organizations.

What it comes down to

A traffic ticket is a flat charge applied to anyone who receives one, regardless of what they earn. The penalties for not paying it are also flat: the same civil assessment, the same suspension, the same reinstatement fee for everyone. What is not flat is the ability to absorb them. For someone with savings, the ticket is a nuisance. For someone without savings, it is the beginning of a problem that can take months to resolve and that touches work, transportation, and debt at the same time.

If you have ever put a ticket in a drawer because there was nothing left in the account to pay it, you were not being irresponsible. You were making the only choice that kept the week from collapsing. The system built the spiral on the assumption that everyone can pay on time. Not everyone can.

The fever starts on a Tuesday night. Wednesday is an eight hour shift at $19 an hour, and there is no paid sick day to cover it.

Staying home costs about $152. That is the whole price of resting, and nobody sends a bill for it. The money is just missing from the next paycheck.

If the fever lasts two days, it is about $304, two fifths of a week’s pay. If a child gets it next, there is another day. Rent does not shrink to match. So the shift gets worked, with the fever, and the cost moves somewhere else.

Who goes without

The Bureau of Labor Statistics surveys employers every March about the benefits they offer. In March 2026, 81 percent of private industry workers had access to paid sick leave. That leaves 19 percent, roughly one in five, who did not.

The gap follows pay. Among workers in the lowest paid quarter of private industry, 61 percent had paid sick leave, so 39 percent went without. Among the highest paid tenth, it was 97 percent.

It follows the type of job too. Part-time workers had access at 59 percent, against 88 percent for full-time workers. Leisure and hospitality, which covers restaurants, hotels and entertainment, was the lowest of any industry at 58 percent. In information and in finance and insurance, it was 98 percent.

The $152 in the opening comes from the same survey. The bureau puts the 25th percentile private industry wage at $18.99 an hour. Eight hours at that rate is $151.92.

Why there is no sick day to take

There is no federal rule that requires one. The Department of Labor says so directly: there are currently no federal legal requirements for paid sick leave.

What federal law does offer is the Family and Medical Leave Act, and it is unpaid. It protects a job for up to 12 weeks during a serious health condition, but only for workers who have been with the employer for at least 12 months, have worked at least 1,250 hours in the past year, and work at a site with 50 or more employees within 75 miles. Someone who just started, works part-time, or works at a small business can fall outside all three. And even for those who qualify, the leave protects the job without replacing the pay.

That leaves it to the states and cities. Fewer than half the states require private employers to provide paid sick leave. In the rest, it depends on the employer.

What going in sick costs

The obvious way to avoid losing $152 is to work the shift. Plenty of people do.

A study of restaurant workers, published in the Journal of Food Protection in 2013 with researchers from the Centers for Disease Control and Prevention, interviewed 491 food workers across nine states. One in five said they had worked at least one shift in the past year while sick with vomiting or diarrhea. Two of the factors tied to that choice were worry about losing the job and worry about leaving coworkers short staffed.

The cost of that choice does not stay with one person. A CDC report in 2023 looked at outbreaks of foodborne illness at restaurants and other food businesses from 2017 to 2019. Sick or infectious food workers contributed to about 40 percent of the outbreaks where a cause was identified. In the same report, fewer than half of the managers interviewed said their business offered paid sick leave to any workers.

So the cost gets paid one way or the other. Either it comes out of the paycheck, or it comes out of the worker’s health and the health of coworkers, customers and family. For a household without savings, the paycheck is the one that cannot give.

The part nobody mentions

The workers least able to lose a day of pay are the ones most likely to lose it.

The Federal Reserve’s survey for 2025 found that 63 percent of adults would cover a surprise $400 expense with cash or its equivalent. The other 37 percent would not. A sick day without pay is a surprise expense, and it lands on the same people who are least likely to have paid leave, because both follow wages.

Going in sick can also make things worse. An illness that might have passed in two days of rest can last longer when it gets worked through, and it can spread to the rest of the household. Then there is the next unpaid day, for the next person who catches it. Every one of those days makes it harder to build the cushion that would have made the first day affordable.

What helps

Two things are real and can be checked.

There may be a state or city law. Because there is no federal requirement, the rules depend on where the job is. A number of states and cities do require private employers to provide paid sick leave, usually earned a little at a time as hours are worked. The state labor department’s website will say whether a law applies, how the time builds up, and what an employer is allowed to ask for. Workers who are covered may already have hours banked that nobody mentioned.

Unpaid leave can still protect the job. For a longer illness, or caring for a family member with a serious health condition, the Family and Medical Leave Act can stop an employer from firing someone for taking the time, as long as the worker meets the length of service, hours and employer size rules. The Department of Labor’s Wage and Hour Division explains the rules and takes complaints. It does not replace the lost pay, but it keeps the job there to come back to.

What it comes down to

A sick day costs nothing for the worker who has paid leave and a full day’s wages for the worker who does not, and the workers without it are mostly the ones paid least. Without a paid day to take, the choice is between losing pay and working sick, and both have a price.

If you have ever gone to work with a fever because you could not afford to stay home, you were not being careless. You were doing the math on a day’s pay, in a system that charges the most for being sick to the people who earn the least.

The boots cost $40 at the discount store in October. They look fine. They feel fine for about a month.

By February the sole has started to peel away at the toe. Water gets in on wet days, and a shift on a concrete floor ends with sore feet and damp socks. By April there is a second $40 pair by the door, because there was $40 that week and there was not $160.

Run that out over four years. Two $40 pairs a year comes to $320. One $160 pair that lasts the same four years comes to $160. The person who could only ever find $40 at a time pays twice as much for boots, and spends those four years in worse ones.

The idea has a name

The novelist Terry Pratchett wrote it into a 1993 book, Men at Arms. A character on a small wage works out that a poor man buying cheap boots over and over spends more across ten years than a rich man who buys one good pair, and still ends up with wet feet. It has been called the boots theory ever since.

It reaches well beyond boots. A cheap mattress that sags in two years. A $20 coat whose zipper gives out in one winter. A cheap vacuum, a cheap phone charger, a set of used tires bought because new ones were out of reach that month. The cheaper item is the one that fits the cash on hand, and it is also the one most likely to need replacing soon.

Why the cheaper pair is the only choice

Most people standing in the shoe aisle can tell which pair is better. What decides it is whether the $160 is there on the day the old pair gives out.

The Federal Reserve’s survey of household finances for 2025 found that 63 percent of adults would cover a surprise $400 expense entirely with cash or its equivalent. That leaves 37 percent who would not. A pair of boots that has to be replaced this week is a surprise expense too, only a smaller one.

The same survey shows how savings split by income. Among adults with family income of $100,000 or more, 75 percent said they had set aside enough to cover three months of expenses. Among adults with family income under $25,000, it was 21 percent.

Without a cushion, the purchase gets made with whatever is in the account that week. That is usually the cheapest version that will do the job for now.

Who pays for work boots

Many jobs run on boots. Warehouse floors, building sites, kitchens, farms and delivery routes all need footwear that holds up, and plenty of them require safety-toe boots.

Federal workplace safety rules say employers must provide required protective equipment at no cost to workers. The same rule then makes exceptions. An employer does not have to pay for ordinary safety-toe boots if workers are allowed to wear them off the job site, and it does not have to pay for normal work boots at all.

So the boots a job depends on are often the worker’s own cost, bought from the same paycheck as rent. When that paycheck is tight, the boots are the $40 pair, and the job wears them out faster than anything else would.

The part nobody mentions

Buying cheap also keeps the person buying cheap.

Every $40 replacement comes out of the money that might have gone toward the $160 pair. The more often the cheap pair fails, the harder it gets to put anything aside, and the further away the better pair moves. The extra cost repeats for as long as money is short, and it is part of the reason money stays short.

It also arrives without a label. Nobody gets a bill marked “extra cost of being broke.” It shows up as a second pair of boots, a second mattress, a second coat, each one a reasonable buy on the day it was made.

What helps

Two things are real and can be checked.

The employer may owe the boots. Under the federal rule on protective equipment, the exception for safety-toe boots only applies when the employer lets workers wear them off the job site. If the job requires specialty protective footwear, or requires safety boots to stay at work, the employer is the one who has to pay. Employers also have to replace protective equipment that wears out, unless the worker lost it or damaged it on purpose. It is worth asking what footwear the job requires, and who is meant to be paying for it.

Boots that fall apart early may still be covered. The Federal Trade Commission explains that most products come with an implied warranty of merchantability, which is the seller’s promise that a product will do what it is supposed to do. It applies even without a written warranty, unless the item was sold “as is” in a state that allows that. In some states implied warranties last as long as four years. A pair of boots whose sole comes away after a few weeks of normal wear is worth taking back to the seller, then to the manufacturer, and if neither helps, to the state consumer protection office.

Neither one puts $160 in an account. What they do is move some of the cost back to where it belongs.

What it comes down to

Cheap things cost more over time because they have to be bought again, and the people buying them are the ones who cannot find the larger sum all at once. The extra cost is small each time, which is why it goes unnoticed, and it keeps coming back for as long as money is tight.

If you have ever bought the $40 boots knowing they would not last, you were not being shortsighted. You were paying for what that week could cover, in a system where being able to pay more upfront is what makes things cheaper.

A client email arrives with a login in it. You want AI to draft the reply, and your cursor is already in the chat box.

Is it safe to use AI at work? For most work, yes, if you check three things before you paste. What is in the text. Whose account you are using. Whether the tool trains on your chats. The text is where people get caught, so that is the part I tested.

I run the day-to-day work at a small web business in Australia. AI is part of how I work every week. Most of what I paste is boring. Some of it is not.

Is it safe to use AI at work?

Mostly. The risk is rarely the AI itself. It is the paste.

A normal work email carries far more than the question you want answered. A phone number in the signature. A login someone sent in a hurry. A line about why a client is late paying.

Australia’s privacy regulator, the OAIC, recommends that organisations do not enter personal information, “and particularly sensitive information”, into publicly available generative AI tools. Health details count as sensitive. So does a lot of what turns up in an ordinary inbox.

So my yes comes with a condition. It is safe when you know what you are pasting. Most people do not look.

What I check before I paste anything

Three questions, in this order.

Is this my account or my employer’s? A work account your employer set up usually comes with its own rules on training and storage. A personal free account does not. If your employer has a policy, it wins.

Is training switched on? In ChatGPT the setting is called “Improve the model for everyone”, under Settings, then Data controls. In Claude it is “Help improve our AI models”, under Settings, then Privacy. On a personal account I turn both off. That stops future chats being used to train models. It does not make the text disappear.

What is in the text? This is the one people skip. It is also the one I could test.

The first test

I did not use real client data for this. That would defeat the point.

Instead I built a sample of the three things I paste most often. A forwarded client email. My own note on invoices. A few lines copied from a developer chat. Every name, number and password in it is made up.

Then I hid 20 things in it that should never reach a chatbot. A client’s mobile number. A WordPress login. A bank account number. A note that one client was late paying because of a divorce. A developer mentioning anxiety. A quote where one client got $3,900 and nobody else was to know the normal price was $5,200.

I gave the whole thing to AI with this prompt:

Before I paste this into an AI tool, check it.
 
[PASTE YOUR TEXT]
 
List everything in it that could identify a person,
a client, or our business, plus any password, key,
money detail or health detail. Quote each one. Then
say what to replace it with. Do not rewrite the text.

It found all 20. It listed 29 items in total, because it also caught things I had not counted, like the clinic owner’s first name tucked inside a testimonial.

“Quote each one” is the line doing the work. Without it you get a vague list of categories. With it you get the exact words to delete.

It also told me something I had not asked. Change the passwords and the payment key anyway, because they had already been sent around in plain text. Cleaning one copy fixes nothing.

A question for you

Think of the last thing you pasted into AI at work. Did you read it from top to bottom first?

I mostly do not.

Can AI check it for you?

Partly. The second test is the one that worried me.

This time I worked the way I would on a busy day. I stripped the obvious things by hand first: names, emails, phone numbers, logins, bank details. Then I left nine things in on purpose, the kind a quick pass misses. A password sitting inside a web address. The client’s domain name in the same address. A surgery mentioned in passing. The divorce. The discount.

Then I ran this:

I have already taken out the names, contact details,
logins and bank numbers from the text below. Assume
I missed things.
 
[PASTE YOUR TEXT]
 
What did I miss? Quote each thing still in there that
names or identifies a person or a business, any login
or key, any health or money detail, and anything a
client would be upset to see. Put the worst first.
Do not tell me what I got right.

It found all nine, plus five more.

First on its list was the password in the web address. Second was the domain name. Its reasoning was blunt. Once the business is named, every gap I had made can be filled straight back in.

It also caught something I had not planted. The client mentioned being away for two weeks. The AI pointed out that this tells a stranger exactly when nobody will be watching the accounts.

“Assume I missed things” is the line to keep. It stops the AI from praising your effort and makes it hunt.

What it got wrong

The testimonial. My sample had a patient back playing footy six weeks after a knee reconstruction, with the physio named.

In the first test the AI flagged it properly. AHPRA says the National Law does not allow testimonials to advertise regulated health services, and comments about symptoms, treatment or outcomes count. That quote should never go on a physio’s homepage.

In the second test, on the same testimonial, it said this:

“Check that the patient agreed in writing to having it on the website.”

Consent does not fix it. The rule applies anyway. Same AI, same words, two different answers, and the second one sounded just as sure as the first.

So I trust it to spot what is in the text. I do not trust it on what the rules say.

The objection I cannot answer

To ask AI what you missed, you have to paste the thing you are worried about.

In my second test the worst item was a live password. The AI found it. It found it because I had already handed it over.

I do not have a clean answer to that. My habit is to strip by hand first, so the AI only sees the leftovers, and to run the check only on an account with training off. But the leftovers are exactly the dangerous ones. You might reasonably decide this last check is better done by hand, slowly, with a coffee.

Where this falls down

• It only checks what you paste. Screenshots, attachments and uploaded files carry the same risks, and I did not test those.
• It cannot see your contracts. Some clients forbid sharing their data with any outside service, names removed or not.
• It is inconsistent on law. The testimonial showed that. Treat any legal point it makes as a lead to check, never a ruling.
• I built the sample, so I knew where everything was. Real text hides things better than mine did.

Skip this if

• Your employer has an approved AI tool and a written policy. Follow that instead.
• You handle health, legal or financial records. Ask before you paste anything, stripped or not.
• You only paste your own writing with no client details in it. You already know what is in there.

Is it safe to use ChatGPT for work emails?

Yes, once the email has been through the three questions. Strip the signature, any login, anything about someone’s health or money, and the name of anyone who has not agreed to it. Then ask for the draft.

What I would tell someone starting

Read the text before the AI does. Most of the risk sits in lines you would never notice in a hurry. A password inside a web address. The reason someone is late paying. One sentence about surgery.

Then use the AI as a second pair of eyes, after your own. Check anything it says about the law yourself.

That client email with the login in it is still sitting in the chat box, waiting for you to press Enter.

What is one thing you have pasted into AI this month that you would rather take back?

The check engine light comes on and the mechanic says $600 — a sensor, maybe a belt, nothing urgent yet. Six hundred dollars might as well be six thousand when rent is due Friday, so the light stays on and life goes on. Three months later the car won’t start on a Tuesday morning, and the $600 fix is now a $2,200 fix, plus a tow, plus three missed shifts because there’s no way to get to work, plus the real risk of losing the job those shifts belonged to.

Nobody drove that outcome on purpose. Every choice along the way — pay rent, buy groceries, keep the lights on — was the right call in the moment it was made. The car just happened to be the thing that could wait, right up until it couldn’t.

Why maintenance turns into emergencies

Cars are supposed to be maintained on a schedule: oil changes, brake pads, tires before they’re bald, belts before they snap. That schedule assumes a person has $150 to $400 sitting around for routine upkeep, on top of everything else. Without that cushion, maintenance stops being scheduled and becomes reactive — nothing gets fixed until it fails, and failure is always more expensive than prevention. A $40 belt replaced on time is a $40 repair. The same belt, ignored until it snaps and takes the water pump with it, is an $800 repair. The car didn’t get more expensive to own. The absence of $40 at the right moment did.

Why the car matters more than the math suggests

For a lot of working people, a car isn’t a convenience, it’s the thing standing between them and their paycheck. Public transit doesn’t reach the job, or doesn’t run the shift hours, or takes two hours each way instead of twenty-five minutes. When the car is down, the paycheck is at risk too — not the cost of the repair, but the cost of the missed shifts stacked on top of it, and eventually the cost of the job itself if the pattern repeats enough times.

That’s the part standard financial advice tends to skip. “Take the bus” assumes a bus goes where the job is. “Buy a more reliable used car” assumes a few thousand dollars in cash or a credit score good enough for reasonable financing — and without either, the realistic alternative is a “buy here, pay here” lot charging 18 to 25 percent interest on a car that breaks down again in a year, restarting the entire cycle.

The insurance and financing penalty on top

Even setting repairs aside, a low credit score or thin credit file raises the cost of owning a car in ways that have nothing to do with the car itself — a subprime auto loan rate that’s double or triple what a good-credit buyer pays for the identical vehicle, or an insurance premium loaded with the same credit-based scoring that affects every other bill. Two people can buy the same used sedan and pay wildly different amounts every month to own it, and the gap has nothing to do with how either of them drives.

What it comes down to

An old car without a financial cushion behind it isn’t just a car. It’s a countdown, and every deferred repair moves the deadline closer while making the eventual bill bigger. The math isn’t a personal failing — it’s what happens when a $40 part and a $2,200 repair are separated by nothing but bad timing and an empty savings account.

If you’ve ever driven around with a warning light on, hoping it would hold until payday, you weren’t being careless. You were making the only decision available to someone who had to choose between fixing a car and paying for everything else that couldn’t wait either.

Payday, and the check in your hand is worth less than the number printed on it. Walk it into a check-cashing store instead of a bank and they’ll take 3 to 4 percent off the top before you see a dollar — $18 to $24 gone on a $600 paycheck, for the privilege of turning a piece of paper into money you already earned.

Multiply that by every two weeks, add in a couple of money orders at $1.50 each because the landlord doesn’t take cash and you don’t have checks, and a person without a bank account can lose $500 or more a year just moving their own money around. Not spending it. Moving it.

How people end up outside the banking system

Nobody chooses this. Roughly one in twenty American households has no bank account at all, and several million more are “underbanked” — technically holding an account but relying on check cashers, money orders, and payday lenders anyway because the account doesn’t actually work for them.

The path in usually starts small: a checking account that dipped negative during a rough month, an overdraft fee that turned into two, a bank that closed the account and reported it to ChexSystems, the banking industry’s version of a credit bureau for checking accounts. Once that report exists, most mainstream banks won’t open a new account for years, regardless of whether the person can otherwise manage money responsibly. The system doesn’t ask what happened. It just remembers the negative balance.

The fee economy that fills the gap

With no bank account, every financial task that would be free or near-free with one becomes a paid transaction. Cashing a paycheck costs a percentage. Paying a bill without a checking account means a money order, a small fee every single time, for every single bill. Prepaid debit cards, marketed as the modern alternative, often carry their own monthly maintenance fees, reload fees, and ATM fees that can rival what a “free” checking account would have cost in overdrafts.

None of these fees are hidden exactly — they’re usually posted on a sign behind the counter. But posted or not, they function as a tax on people who’ve already been pushed out of the system that would have let them avoid the tax in the first place. The person paying 4 percent to cash a check is, almost by definition, the person who can least afford to lose 4 percent of it.

The catch-22 nobody mentions

The cruelest part is how self-reinforcing it is. A negative balance from years ago keeps someone locked out of a free checking account today, which pushes them toward check cashers and prepaid cards, which cost more than a checking account ever would have, which makes it that much harder to build the cushion that would have prevented the original overdraft. The system that punished one bad month keeps charging for it indefinitely, through a completely different set of fees, at a completely different set of counters.

Some credit unions and community banks offer “second chance” checking accounts built for exactly this situation, and they’re worth asking about — but they’re not advertised the way overdraft-happy big banks advertise their own products, so most people paying the check-cashing tax have never heard they exist.

What it comes down to

Not having a bank account isn’t free. It’s one of the most expensive ways to handle money there is, and it’s disproportionately forced on the people who have the least room to absorb the cost.

If you’ve ever paid to cash your own paycheck, or bought a money order because it was the only way to pay a bill, you weren’t being irresponsible with money. You were paying admission to a financial system that had already locked its front door behind you — and charging you every time you walked back in through the side.

A friend of mine took a promotion two years ago — same job, better title, an extra $1.75 an hour. She did the math on a napkin before she said yes: more hours at a slightly better rate should have meant a few hundred extra dollars a month. Instead, her next paycheck came with less breathing room than before, not more. Her income had crossed a line on a form somewhere, and on the other side of that line, she lost most of her childcare subsidy in the same month her raise kicked in.

She didn’t do anything wrong. She ran straight into what people who study this call a benefits cliff, and once you know what to look for, it explains a pattern a lot of working people have lived through without ever having a name for it.

What a cliff actually looks like

Most people assume support like SNAP, Medicaid, childcare assistance, or housing vouchers phases out gradually as income rises — earn a little more, lose a little help, still come out ahead overall. Some programs do work that way. Plenty don’t. A number of these benefits are structured with hard income thresholds: cross it by even a dollar and the assistance doesn’t taper, it drops off a ledge, sometimes all at once.

Childcare assistance is one of the sharpest examples. A single parent working close to a program’s income limit can lose a subsidy worth $800 or $1,000 a month over a raise worth a fraction of that. Medicaid works similarly in states that haven’t smoothed the transition — cross the income line and a family can lose health coverage for kids or a spouse well before a new job’s insurance kicks in, if it ever does. SNAP has its own thresholds. So do some housing vouchers. None of them coordinate with each other, so a raise that clears one program’s limit can also clear two others in the same month, stacking losses a person never sees coming until the next bill arrives.

Why this isn’t a design flaw anyone intended, but is one anyway

Benefits cliffs weren’t built to punish people for working more. They’re what happens when dozens of separate programs, each with its own income rules written by a different agency in a different decade, get stacked on top of each other with no one responsible for checking whether the combination makes sense for a real person’s paycheck. Economists across the political spectrum agree the cliffs are a problem. Fixing them means redesigning how multiple federal and state programs interact, which is a slower, less visible fight than almost anything else in the debate about poverty.

In the meantime, the person standing at the edge of the cliff is the one who absorbs the miscalculation — not the agencies, not the employer offering the raise, not the lawmakers who wrote the thresholds. Her.

The choice nobody should have to make

Once people learn where the cliffs sit, some make a rational, defensible decision that looks irrational from the outside: turning down extra hours, asking to stay just under a threshold, or hesitating before accepting a raise or promotion. It gets read as a lack of ambition. It’s the opposite — it’s someone doing the arithmetic more carefully than the system that put them in this position ever did, and choosing not to work harder for less money.

If you’ve ever quietly turned down more hours, or felt a flash of dread instead of relief when a raise came through, you weren’t being ungrateful and you weren’t being lazy. You were looking at a cliff that the paperwork never warned you was there, and doing the only sane thing available: not walking off it.

Nobody sits you down and explains what a low credit score actually costs. You find out sideways — a landlord asks for triple the deposit instead of one month’s rent, a cell phone carrier tells you that plan requires a $400 “assurance” payment upfront, or your car insurance renewal shows up $60 a month higher than your coworker’s, even though you’ve never filed a claim and he got a speeding ticket last spring.

None of that shows up on the loan paperwork where you’d expect it. It shows up everywhere else instead.

The deposits nobody budgets for

Good credit lets you skip lines that bad credit forces you to stand in, in cash, upfront. A one-bedroom apartment might ask for first month, last month, and a security deposit from anyone — but with a low score, some landlords tack on an extra half or full month “risk deposit” on top. Utility companies do the same thing: show up with a thin or damaged credit file and the electric company wants $150 to $300 in your hand before they’ll turn the power on, money someone with a 720 score never has to produce. Cell carriers run the same check. So do some employers, on background screenings that quietly weigh a credit report they’ll never admit factored into the decision.

Add it up and the “bad credit tax” on simply setting up a normal life — an apartment, power, a phone, sometimes a job — can run into the thousands, due all at once, at the exact moment someone can least afford a thousand-dollar hit: right after a move, a layoff, or a medical bill that wrecked the score in the first place.

The insurance number nobody explains

In most states, auto and home insurers are allowed to factor a “credit-based insurance score” into your premium — a different number than your regular credit score, but built from the same raw material: how you’ve handled debt. Two neighbors with identical cars, identical driving records, identical coverage can pay premiums hundreds of dollars apart a year, and the gap has nothing to do with how either of them drives. It has to do with a hospital bill one of them is still paying off.

Insurers will tell you the math holds up statistically across large pools of people. Maybe it does. It doesn’t change what it feels like from the inside: getting charged more to protect the same car, because a system decided your unpaid ER visit from two years ago says something about your risk behind the wheel.

How people actually end up here

The credit score conversation usually assumes recklessness — maxed cards, blown budgets, missed payments out of carelessness. For a lot of people the real story is one bad month that never fully unwound: a layoff that ran two weeks longer than the emergency fund lasted, a car repair put on a card because the alternative was losing the job that depended on driving to it, a medical bill that landed in collections while the person was still working, just not fast enough to cover a $4,000 ambulance ride.

The score doesn’t know the difference between reckless and unlucky. It just remembers the missed payment, and every deposit, premium, and rate for years afterward remembers it too.

What it comes down to

A low credit score doesn’t just make borrowing more expensive. It makes ordinary life more expensive — the roof, the lights, the phone, the insurance you’re required by law to carry. And it charges the most, in cash, at the moments people already have the least of it.

If you’ve ever paid a deposit or a premium that felt wildly out of proportion to anything you actually did, you weren’t imagining it. You were paying for a number that doesn’t know your story, only your worst month — and charging you for it, over and over, long after that month ended.

There’s a version of poverty nobody puts in the budgeting spreadsheets: the version where being short on money actively costs you more money.

You know it if you’ve lived it. The $35 overdraft fee on a $6 coffee because your paycheck landed a day late. The $400 tire that turns into an $800 tow, a rental car, and a missed shift because you couldn’t afford the $400 three weeks earlier when the tread was already going. The payday loan that felt like the only option at 9pm on a Tuesday with the power bill due Thursday.

None of this is a personal failing. It’s arithmetic. And almost nobody who gives financial advice for a living has actually run the numbers on it.

The tax nobody votes on

Economists have a name for it — the “poverty premium” — but you don’t need the term to know the pattern. Poor people pay more for almost everything that matters:

More for banking, because a “free” checking account isn’t free once you’ve been hit with three overdraft fees in a bad month. More for credit, because a low credit score doesn’t stop you from needing a car loan — it just makes the loan cost more. More for appliances, because a $300 washing machine that lasts two years is a worse deal than the $700 one that lasts twelve, and you don’t have $700. More for groceries, if the only store in walking distance is the corner shop instead of the discount supermarket across town you can’t get to without a car you also can’t afford.

Every one of these is a small, individually defensible decision. Stacked together, they’re a tax on not having a cushion — paid by the people who can least afford it, straight to the people who can.

The advice that doesn’t survive contact with real life

This is why so much popular financial advice lands wrong. “Just build an emergency fund” assumes you have income left over after rent, food, and the minimum payments that are already eating you alive. “Buy in bulk to save money” assumes you have $150 free on a Tuesday and a car to haul it home in, not $40 and a bus pass. “Cut the coffee, cut the streaming” was never the math problem — rent was.

None of it is malicious, most of it, but it’s written by people several rungs up the ladder, aimed down at people who are already doing the math better than the advice-giver, out of necessity, every single day.

What actually helps

Not a five-step plan. A shift in where you look for the leak.

The overdraft fees, the payday loans, the late fees, the “convenience” charges — those are usually the biggest, most fixable drain, bigger than the latte, bigger than the subscription nobody mentions canceling. A single free or low-fee account at a credit union instead of a big bank. A grace-period reminder set the day before, not the day of. Ten minutes on the phone asking a biller to waive one fee, once, because most will if you just ask.

None of it fixes the underlying problem, which is that wages haven’t kept pace with rent in most of the country for two decades. But it stops some of the bleeding while that bigger problem sits, unsolved, above everyone’s pay grade.

If you’ve felt like you were bad at money for paying fees everyone else seems to avoid — you weren’t bad at money. You were paying the tax that comes with not having enough of it, and almost nobody bothered to tell you the tax existed.