She cashes her paycheck at the storefront on the corner. A thousand dollars. The fee is 2.34 percent \u2014 $23.40 \u2014 gone before she’s bought a single thing. Then a money order for rent: $2. That’s $25.40 extracted for the act of receiving wages she already earned and converting them into a form her landlord will accept.<\/p>\n\n\n\n
Two weeks later: the car goes. The repair is $400. She doesn’t have $400. What she has is access to a payday lender, who will advance her $400 at $15 per $100 per two-week period \u2014 the industry standard, what annualizes to 391 percent APR. The fee: $60. She can’t pay it off in two weeks \u2014 if she could, she wouldn’t have needed the loan. She rolls it over. Another $60. The $400 emergency has now generated $120 in fees \u2014 and she’s still carrying the principal.<\/p>\n\n\n\n
Consider the same emergency for someone who keeps a credit card with a 20 percent APR. Four equal monthly payments of roughly $104. Total interest charged, on a declining-balance calculation: about $17. Total cost of the emergency: $417.<\/p>\n\n\n\n
The only variable is how much money the person started with. One of them pays $17 for a $400 problem. The other pays $120 in fees alone \u2014 and counting, with the principal still owed.<\/p>\n\n\n\n
Money has a price. What the data shows is that the price of money varies by a factor that makes any other consumer pricing variation look modest.<\/p>\n\n\n\n
At the top of the credit spectrum: the 30-year fixed mortgage. As of April 2026, the Freddie Mac Primary Mortgage Market Survey places the average rate at 6.30 percent. This is how people with stable income, substantial assets, good credit history, and the time to shop multiple lenders borrow money for the largest purchase of their lives.<\/p>\n\n\n\n
A tier down: credit cards. The CFPB’s 2025 Consumer Credit Card Market Report found the average APR for general-purpose cards reached 25.2 percent in 2024 \u2014 itself a record high, driven partly by the rate environment of recent years. Subprime personal loans cluster between 36 and 60 percent depending on the lender. These are the products available to people with imperfect credit histories: higher cost, shorter terms, less flexibility.<\/p>\n\n\n\n
Then payday loans. The CFPB’s figure, drawn from its comprehensive study of storefront payday lending, is 391 percent APR \u2014 what you get when a $15-per-$100 fee on a two-week loan is annualized. This isn’t a fringe product. It’s the standard pricing structure of a legal, regulated, mainstream industry operating in most US states.<\/p>\n\n\n\n
The rent-to-own market makes payday lending look conservative. A washing machine with a retail price of $500, structured as a rent-to-own agreement at $25 per week over 78 weeks, costs $1,950 total. The effective annual cost of that financing, calculated as a rate, runs well over 100 percent \u2014 often far higher. Most states classify rent-to-own as a lease rather than a credit transaction, which conveniently exempts it from the usury laws that would otherwise cap it.<\/p>\n\n\n\n
The standard explanation for this gradient is risk. Mortgage lenders accept collateral, underwrite carefully, lend to creditworthy borrowers. Payday lenders lend to anyone with a paycheck and a bank account, with no credit check, to people with documented histories of financial stress. Of course the rate is higher. The risk is higher.<\/p>\n\n\n\n
This is true as far as it goes. It doesn’t go as far as people think.<\/p>\n\n\n\n
What the rate gradient actually measures \u2014 more than risk \u2014 is the borrower’s ability to wait, to shop, to walk away. A borrower with good credit, stable income, and a substantial savings cushion can take three months comparing mortgage offers, walk away from a rate they don’t like, wait for better conditions. That leverage is worth 6 percent. A borrower with no savings buffer, a car that needs fixing by Monday, and no credit card to fall back on is not in a position to shop. That powerlessness costs 391 percent. The credit risk is real; the pricing differential is also, substantially, a function of bargaining position.<\/p>\n\n\n\n
State regulation: the same product, different law<\/strong>\n\nColorado voters passed Proposition 111 in November 2018 by 77 percent of the vote, capping payday loan APRs at 36 percent. Before the cap, the average APR on payday loans in Colorado was 129 percent, per the Colorado attorney general's office. The law took effect February 1, 2019.\nIdaho imposes no cap on payday loan interest. APRs in Idaho have reached 652 percent, per Center for Responsible Lending documentation of rates advertised by major payday chains.\nSame product. Same industry. Different legislature. The gap between 36 percent and 652 percent is not a market outcome. It is a legislative choice, made by specific people, maintained by other specific people choosing not to change it.<\/code><\/pre>\n\n\n\nThe premium on everything<\/h3>\n\n\n\n
If the poverty premium existed only in credit markets, you could construct a coherent argument that it reflects lending risk. But it doesn’t exist only in credit markets. It exists in insurance, in energy, in financial services access, in food. When the same structural pattern appears across industries with fundamentally different cost structures, something more systematic is at work.<\/p>\n\n\n\n
The foundational UK study comes from the Personal Finance Research Centre at the University of Bristol, published in 2016. Their headline figure: the average poverty premium \u2014 the additional annual cost paid by low-income households compared to better-off households for equivalent goods and services \u2014 is \u00a3490 per year per household. Car insurance was the largest single component: some low-income households paid nearly \u00a3300 more per year, with the pricing driven primarily by geographic location rather than individual driving history. The insurer isn’t pricing your driving. It’s pricing your postcode. Where you can afford to live determines what you pay to insure a car. An additional \u00a3160 per year falls on households paying car insurance monthly rather than annually \u2014 because monthly is what you do when you don’t have the cash to pay the full annual premium at once.<\/p>\n\n\n\n
The combined cost of these penalties, across essential markets, is not marginal. It is the equivalent of weeks of income.<\/p>\n\n\n\n
The UK energy market illustrates the mechanism most cleanly. Households that couldn’t commit to monthly direct debit payments \u2014 the payment method banks require, which requires a functioning bank account and the ability to maintain a minimum balance \u2014 were placed on prepayment meters, which charged higher per-unit rates. The price of being unable to commit to monthly payments was a higher price per unit of energy consumed. The UK energy regulator Ofgem scrapped the prepayment meter premium in 2023. Before that, it was simply the cost of being poor in the UK energy market. The mechanism \u2014 pay more for the less favorable payment method \u2014 is structurally identical to what happens in insurance, in financial services, in every market where the cheaper option requires committing resources in advance.<\/p>\n\n\n\n
In the United States: the FDIC’s 2023 National Survey of Unbanked and Underbanked Households found 4.2 percent of US households \u2014 approximately 5.6 million \u2014 unbanked. For those households, the alternative to a bank account is a sequence of transactions, each with its own cost. A 1997 Consumer Federation of America survey of 111 outlets across 23 cities documented check-cashing fees averaging 2.34 percent of the check value. A money order at USPS costs $1.25 to $2; private retailers charge more. A worker receiving biweekly paychecks who cashes each one at 2.34 percent and pays two money orders per month for rent and utilities is paying several hundred dollars per year for the privilege of accessing wages already earned. The bank account that would eliminate these costs requires a minimum balance they may not be able to maintain and overdraft fees that can turn a $5 shortfall into a $35 penalty.<\/p>\n\n\n\n
Food completes the picture. A 2018 USDA Economic Research Service study found variety-adjusted food prices 3.5 percent higher in low-access urban census tracts than in otherwise similar tracts with better access \u2014 a gap that rises to 9.2 percent for households who cannot shop outside their census tract. More recently, a 2025 study by researchers at the University of Florida (Grigsby-Calage et al., American Journal of Agricultural Economics) found that in urban census block groups with only one grocery store \u2014 roughly 14 percent of such block groups \u2014 dollar store openings were associated with measurable grocery store losses and reduced food access. The mechanism: dollar stores are cheaper to operate and can undercut on price for a limited product range while accelerating the exit of full-service grocers from marginal markets.<\/p>\n\n\n\n
Bristol University methodology<\/strong>\n\nThe \u00a3490 figure is a weighted average \u2014 it reflects both the size of the premium in each market and the proportion of low-income households actually incurring it. This methodology, compared with earlier estimates as high as \u00a31,300 per year, is more conservative. The true figure for households facing multiple simultaneous premiums is substantially higher than the average.\nThe UK prepayment meter energy premium has been scrapped as a regulatory matter since 2023. The mechanism \u2014 pay more because you can't pay in the preferred format \u2014 remains structurally active in every market where the cheaper option requires upfront commitment.<\/code><\/pre>\n\n\n\nThe car you cannot afford not to have<\/h3>\n\n\n\n
The markets documented above all operate within a physical context. In the United States, that context is car-shaped.<\/p>\n\n\n\n
American cities were built around the assumption of car ownership as a baseline. That assumption is embedded in zoning codes, in the distance between residential areas and commercial centers, in the routing of bus lines and the scheduling of transit. For the bottom two income quintiles, the ITDP’s 2024 report on transportation costs in the United States documents that these households spend more than 25 percent of their budgets on transportation in car-dependent areas. Federal data from the Department of Transportation shows the lowest-income households spent 32 percent of before-tax income on transportation in 2023 \u2014 more than twice the share of higher-income households in absolute budget terms.<\/p>\n\n\n\n
Without a car in a car-dependent city, the arithmetic of daily life becomes punishing. The nearest full-service grocery store may require two or three bus connections and sixty to ninety minutes of travel time each way. Medical appointments become logistical undertakings significant enough that people defer them. The job market contracts to whatever is reachable by transit \u2014 which in most American cities is a fraction of total employment, skewed toward certain industries and neighborhoods.<\/p>\n\n\n\n
When a low-income household does own a car, it typically owns an older, cheaper vehicle. An older vehicle means higher expected repair costs, worse fuel economy, worse emissions, and a higher probability of catastrophic mechanical failure at an inconvenient moment. It also means higher insurance rates in some markets, because older vehicles attract different underwriting assumptions. The people for whom a major car repair is most financially devastating are also the people most likely to face one.<\/p>\n\n\n\n
This is the paradox that infrastructure enforces: the people for whom car ownership is most financially precarious are the people for whom not owning a car is most costly. Both horns of the dilemma are expensive. There’s no cheap option. There’s just the choice of which kind of expensive to accept.<\/p>\n\n\n\n
The time cost is real. Transit-dependent commutes in car-dependent cities routinely run an hour or more longer than their driving equivalents \u2014 compounding, over a working year, to hundreds of hours that can’t be reclaimed for additional work, family, or rest. It doesn’t appear on any household budget sheet. It disappears.<\/p>\n\n\n\n
What the diaries say<\/h3>\n\n\n\n
Three sections of structural evidence, and a natural next move would be to conclude that people at the bottom of the income distribution are economically passive \u2014 unable to navigate markets that are systematically tilted against them, overwhelmed by the complexity, defaulting to bad decisions.<\/p>\n\n\n\n
The financial diaries don’t support that reading.<\/p>\n\n\n\n
Daryl Collins, Jonathan Morduch, Stuart Rutherford, and Orlanda Ruthven spent a decade tracking every financial transaction made by households living on under $2 per person per day in Bangladesh, India, and South Africa. More than 250 financial diaries, recorded in real time, penny by penny. The resulting book, Portfolios of the Poor (Princeton University Press, 2009), is one of the most methodologically thorough studies of how poor households actually manage money.<\/p>\n\n\n\n
What they found: these households are not passive. They are managing complex, layered financial portfolios \u2014 simultaneously. A single household might hold a small savings account for emergencies, belong to a rotating savings club (a ROSCA, where members pool contributions and each takes a turn receiving the full pot), carry an active microfinance loan used for productive investment, maintain an informal tab with a trusted shopkeeper for day-to-day smoothing, and hold obligations to extended family that function as a reciprocal insurance system. Each instrument serves a distinct liquidity function. Each relationship requires maintenance and trust. The whole structure is more sophisticated than what most middle-class households in wealthy countries manage. The margin for error is zero \u2014 and so the financial management has to be.<\/p>\n\n\n\n
Not because these households are financially exceptional. Because the alternative to active, sophisticated management is ruin.<\/p>\n\n\n\n
The US domestic complement comes from Jonathan Morduch and Rachel Schneider’s The Financial Diaries (Princeton University Press, 2017), which tracked 235 low-to-moderate-income American households through a full year of financial transactions. The central finding: income volatility, not just income level, is the defining challenge. Households in the study experienced an average of 2.2 months per year when income spiked more than 25 percent above their monthly average and 2.4 months when it fell more than 25 percent below. Nearly five months per year when income looked nothing like the monthly baseline. Long-term saving \u2014 the canonical advice for financial security \u2014 is mechanically difficult when you cannot reliably predict next month’s income. It is not a failure of will. It is a failure of the income structure to support the savings structure being prescribed.<\/p>\n\n\n\n
The population that popular discourse describes as financially imprudent is, in the data, often the most financially attentive. Not by choice. Because the alternative is ruin.<\/p>\n\n\n\n
The bandwidth tax<\/h3>\n\n\n\n
The sophistication documented in the financial diaries is real. But it is not free. Managing money at zero margin \u2014 tracking every obligation, calculating every trade-off, maintaining every relationship \u2014 requires exactly the kind of sustained executive function that financial scarcity directly depletes. Both things are simultaneously true.<\/p>\n\n\n\n
The foundational evidence is Anandi Mani, Sendhil Mullainathan, Eldar Shafir, and Jiaying Zhao’s 2013 paper “Poverty Impedes Cognitive Function,” published in Science (Vol. 341, pp. 976-980; DOI: 10.1126\/science.1238041). The paper reports two studies.<\/p>\n\n\n\n
In the first, researchers approached shoppers at a New Jersey mall and presented them with a financial scenario \u2014 how would you handle a car repair costing either $150 or $1,500? They then measured cognitive performance on two unrelated tasks: Raven’s Progressive Matrices, a standard measure of fluid intelligence, and a spatial compatibility task. Low-income participants showed significantly reduced performance when the scenario involved the larger sum. The same people, presented with the smaller sum, performed normally. Higher-income participants showed no such effect in either condition. The financial scenario consumed cognitive resources \u2014 specifically for people to whom the scenario was financially meaningful.<\/p>\n\n\n\n
The second study is methodologically stronger, because it uses within-subject design and eliminates most alternative explanations. Tamil Nadu sugarcane farmers receive approximately 60 percent of their annual income in a single harvest. The same individuals were tested before and after harvest \u2014 poor before, flush after. Pre-harvest scores were measurably lower on cognitive assessments. Post-harvest scores were measurably higher. The effect size: equivalent to approximately 13 IQ points, or losing a full night’s sleep. The variable is conditions, not people. The same person, with the same brain, in different financial circumstances, shows different cognitive capacity.<\/p>\n\n\n\n
The mechanism is what Mullainathan and Shafir call bandwidth \u2014 the finite cognitive resource that handles attention, executive function, and self-control. Financial scarcity forces constant mental computation: what can be paid, what deferred, what the consequences of each choice are, how competing obligations rank. This background processing consumes bandwidth that then cannot be applied to other tasks.<\/p>\n\n\n\n
Long-term financial planning \u2014 the activity most commonly prescribed as the solution to poverty \u2014 is precisely the kind of deliberate, executive-function-heavy thinking that financial scarcity makes hardest to sustain. Managing a payday loan rollover and engaging in retirement planning are competing demands on the same cognitive resource. And poverty, structurally, ensures the emergency always wins. The planning that would theoretically reduce future emergencies cannot be done by a mind fully occupied with the current one.<\/p>\n\n\n\n
The 2013 Science paper attracted scrutiny. A 2013 comment by Wicherts and Scholten (DOI: 10.1126\/science.1246680) raised concerns about the mall study’s design, and some replication attempts of related effects have had mixed results. The farmer study \u2014 a within-subjects natural experiment with strong controls for individual differences \u2014 sits on firmer ground. The core finding should be treated as meaningful but contested, not as an unassailable law.<\/p>\n\n\n\n
Decision fatigue and the broader mechanism<\/strong>\n\nRoy Baumeister's ego depletion research offers a psychological mechanism for how cognitive load produces these effects: self-regulation draws on a limited resource that depletes with use. The connection to Mullainathan and Shafir's scarcity framework is intellectually direct \u2014 financial scarcity forces continuous self-regulatory effort, depleting the resource needed for further self-regulation.\nEgo depletion research also carries a significant replication controversy of its own. The mechanistic account is plausible and fits the Mani et al. findings; it is not proven to the standard of the farmer study.<\/code><\/pre>\n\n\n\nDesigned, not accidental<\/h3>\n\n\n\n
Five sections of evidence; one question still unasked. Not whether the poverty premium exists \u2014 the data settled that \u2014 but who maintains it.<\/p>\n\n\n\n
Markets do something correctly: they price risk. A borrower who defaults more often should face higher rates. An insured vehicle parked in a neighborhood with more accidents should carry a higher premium. A food retailer operating in a market with lower volume and higher shrinkage should charge higher prices. Each of these individual pricing decisions is locally rational. Each has a coherent actuarial justification.<\/p>\n\n\n\n
Taken together: a system that charges the poorest participants the highest rates for the necessities of life, in every market they enter, at every transaction. The logic that justifies each individual premium also produces their aggregate effect.<\/p>\n\n\n\n
But there are two distinct kinds of design operating here, and they’re worth keeping separate.<\/p>\n\n\n\n
The first is deliberate construction. Payday loan rates in Idaho are 652 percent because Idaho’s legislature chose not to cap them. Colorado’s rates dropped to 36 percent because Colorado voters chose otherwise. Rent-to-own APRs are legal wherever no statute limits them \u2014 and most states don’t. These are legal structures created and maintained by specific actors. The administrative complexity of SNAP eligibility verification, EITC documentation, housing assistance applications, and Medicaid re-enrollment imposes compliance costs \u2014 time, documentation, repeated interactions with bureaucracies \u2014 that fall most heavily on the people with the least capacity to absorb them. Pamela Herd and Donald Moynihan’s Administrative Burden: Policymaking by Other Means (Russell Sage Foundation, 2018) makes the core argument: this complexity is often deliberately imposed by actors who want to restrict access to benefits without appearing to have eliminated them. The EITC is a useful case. The IRS estimates approximately 20 percent of eligible households fail to claim it each year \u2014 roughly 19 percent to 22 percent depending on year, representing billions in unclaimed benefits. Among the primary reasons: administrative difficulty, complexity, and fear of audits. When administrative simplification has been introduced \u2014 simpler forms, pre-filled returns, outreach programs \u2014 take-up increases. The complexity is not inevitable. It is maintained.<\/p>\n\n\n\n
The second kind of design is deliberate maintenance through inaction. Zip-code insurance pricing is actuarially grounded \u2014 neighborhoods do generate different claim rates, and insurance companies didn’t design the neighborhood. But regulators have consistently declined to ban geographic pricing, which means the decision to permit it is renewed with each legislative session that passes without addressing it. Car-dependent urban infrastructure reflects decades of investment decisions \u2014 not a conspiracy against transit commuters, but one that stagnant public transit funding and continued highway expansion actively maintains. Food deserts emerge from private grocery chain exit decisions, but zoning and land-use rules that permit dollar stores while imposing barriers on grocery store development have been written and renewed. The choice to leave these structures in place is itself a choice.<\/p>\n\n\n\n
The article’s verdict is not that someone designed the poverty premium in a room with a blueprint. It is that the poverty premium is a functional outcome \u2014 measurable, systematic, operating across markets \u2014 that is politically produced and politically maintained. People with the power to change it have, market by market and program by program, chosen not to. That choice has a price. It’s just that someone else pays it.<\/p>\n\n\n\n
She’s back at the check-cashing window. The $23.40 fee looks, from here, different than it did at the top of the page. It’s not an anomaly. It’s one data point in a pricing architecture that runs across credit, insurance, energy, food, transportation, and the bureaucratic systems built to provide relief from these markets \u2014 each component of which charges the same population extra for access that others receive at a discount or for free.<\/p>\n\n\n\n
What does it mean to maintain a system that charges its most vulnerable participants the highest rates for the necessities of life, requires them to sustain the most complex financial management in the economy, taxes the cognitive bandwidth they’d need to plan their way out, and then asks why they can’t seem to get ahead?<\/p>\n\n\n\n
Don’t answer that too quickly.<\/p>\n\n\n\n