What this covers

Patrick McKenzie and Russ Roberts examine credit card economics in depth, with McKenzie arguing that cards function as an infrastructural achievement in financial inclusion and risk transfer, not a regressive subsidy scheme. The conversation maps the mechanics of card revenue—interchange fees, net interest income, and ancillary charges—and disputes the widely circulated claim that poor cash users effectively fund rich rewards-card users. McKenzie contends that high-income cardholders not only earn the bulk of rewards but also generate the bulk of total spending and interchange income, making the arithmetic of the subsidy claim fail on its face. He addresses the empirical evidence cited to support the opposite view, citing Federal Reserve research by Sumit Agarwal that actually finds redistribution flowing from less-savvy to more-savvy consumers rather than cleanly from poor to rich.

The discussion ranges across how cards reallocated credit risk from retailers to banks—a shift that reduced retailer fragility and replaced social-capital-based underwriting with data-driven risk scoring—and how different FICO bands create different profitability curves for issuers. McKenzie explores the consequences of policy interventions: usury caps and interchange fee caps that push marginal borrowers out of formal credit into payday loans, social borrowing, or deferred utilities, costs that lack visible APR labels yet often exceed the rates they were designed to prevent. He frames card rewards as a form of price discrimination comparable to couponing, describes the optimal rate of fraud as deliberately non-zero, and notes that even non-revolving cardholders receive credit through grace periods. The analysis invokes the Coase Theorem to show how liability assignment shapes incentive structures, and touches on how trust levels in a society determine fraud rates and transactional friction.

Sharpest takeaway

McKenzie argues that credit cards are an infrastructural triumph that expands financial inclusion and transfers risk efficiently, and that the popular claim that poor cash users subsidize rich rewards-card users is fundamentally false because high-status customers generate the bulk of spending and interchange income.

  • Credit cards make money primarily through interchange and net interest income, with rewards funded by interchange on a per-portfolio basis, not by interest payments from the poor.
  • Better-off consumers spend a disproportionate share of total purchases, so they cannot be net-subsidized by lower-spending cash users.
  • Interest rate caps and interchange caps push marginal borrowers out of formal credit into worse alternatives, with costs that are less visible but real.

The argument · threads3 threads · 29 claims
0.85

Liability assignment determines merchant and bank incentives to prevent fraud and loss.

1 pointcentrality 5/5
  • The allocation of fraud and chargeback liability between merchant and bank is an application of the Coase Theorem: the assignment of property rights (who bears the loss) sets the incentive structure, so that under the current configuration the merchant has the strong incentive to avoid fraud while the bank, whose customer is the cardholder, has less. Changing that assignment would shift incentives accordingly.

    “this is an application of what I think of as the Coase Theorem, which is basically property rights. The assignment of property rights sets in motion a set of incentives. And if you change the property rights, you're going to have different incentives on the parties.”

0.85

Card issuer bundles replace costly in-house credit with computerized, professional, and diversified risk management.

1 pointcentrality 5/5
  • Credit card issuers offer merchants a bundle that replaces in-house store credit: computerized math instead of bookkeepers, professional collections departments instead of salespeople, cheap deposit funding instead of expensive working capital, adequate capitalization against losses instead of thin equity, and diversification across regions and sectors instead of concentrated local risk — making credit provision far more efficient than the old store-ledger model.

    “They'd have access to cheap deposits to fund loans, rather than expensive working capital. They'd be adequately capitalized against losses, rather than having tiny margins backed by almost no equity, like most retailers. They'd diversify against regional and sectoral risk”

0.44

Merchant pricing structures reflect fixed transaction costs and variable non-payment risk.

1 pointcentrality 2/5
  • Merchants impose minimum purchase amounts for card use partly because of the fixed component of the transaction cost, and pricing uses a percentage plus fixed fee because non-payment loss scales with transaction size.

    “if you don't pay, it's a bigger loss for a big transaction than if you don't pay for a smaller transaction. So, that would be why you'd want to have a percentage rather than a fixed cost at all.”