I see way too many Dave Ramsey videos online…

His advice seems entirely hokey. Use cash envelopes. Grocery budget in one envelope, entertainment in another. When the envelope is empty, stop spending. Pay off your smallest debt first, regardless of interest rate. Build momentum. Feel the wins.

There’s one type of economist (me, that’s me), who winces at this. Money is fungible—a dollar in one envelope is worth the same as a dollar in another. Paying low-interest debt before high-interest debt leaves money on the table. These rules violate basic principles of optimization.

There’s another type of economist (me, also me, when I’m being a better economist), who says wait. Millions of people swear by the envelopes. And it’s not just Ramsey. Rules of thumb are everywhere: round down small purchases, treat sunk costs as sunk, set aside money before you see it. Personal finance is full of these heuristics that look wrong to an optimizer but seem to be popular in practice.

So let’s flip it around: what constraints make this behavior sensible? What problem does it solve? Why do people rely on heuristics when making decisions?

A core of economics is about people maximizing subject to constraints. Usually, economists focus on budget constraints. Sometimes we add time constraints. But there’s another constraint that price theorists have long taken seriously: cognitive costs.

Decision-making takes effort. Comparing options requires attention. Tracking spending causes fatigue. This is a huge part of Thomas Sowell’s Knowledge and Decisions, which we are always mentioning here.

Deliberate decision making is not a free good; that is why there are thermostats and payroll deductions. Decision making has costs, including time, stress, fatigue, insomnia, and heart attacks. Clearly, it is something that must be economized.

Armen Alchian made similar points about how market institutions economize on information costs. The insight isn’t new.

How do we economize on decision-making? Rules of thumb. Mental shortcuts. Categories that reduce complex comparisons to simple ones. Instead of evaluating every dollar against every possible use, we create buckets: “entertainment,” “clothing,” “eating out,” and optimize within them.

Think about what “rational” budgeting would require. Every dollar has an opportunity cost—the best alternative use across your entire lifetime. We introduce opportunity cost as some simple idea in Econ 101 but taking it literally in some omniscient sense would be crazy. To optimize perfectly, you’d need to compare this jacket against next year’s vacation, your retirement savings, and every other possible expenditure. Nobody does this. Nobody can.

We can see this made explicit in AI systems, where complicated reasoning literally costs more tokens, and therefore more money. Human cognition isn’t priced so precisely, but the constraint is just as real.

Someone who puts grocery money in one envelope and entertainment money in another probably understands fungibility just fine. They know they can move the dollars between envelopes, and probably have many times, breaking their own rules.

But the envelope system reduces the cognitive cost of evaluating purchases. Mental accounting—separating money into categories like “entertainment” or “clothing”—creates a simpler optimization problem than tracking every dollar against every possible use.

Cognitive costs can be modeled as real constraints, with testable implications. In my work with Mark Whitmeyer, we explicitly add information acquisition costs to a model of consumer choice.

This is just bread-and-butter constrained maximization, with a twist. The constraint happens to be monetary plus cognitive rather than just monetary.

Once we recognize cognitive costs as constraints, we can ask how institutions affect them. Some institutions raise cognitive costs; others lower them.

Markets, through prices, economize on information. They compress enormous amounts of information into a single number. When you see a price of $50 for a jacket, you don’t need to know anything about the supply chain, the cost of materials, the wages of workers, or the scarcity of retail space. The price summarizes all of that for you.

Friedrich Hayek made this point about the price system as a whole. No one needs to know why copper is scarce—whether a mine collapsed or demand surged in China. The price increase alone tells people to economize. Prices allow coordination without requiring that anyone understand the full picture.

At the individual level, this is the idea that prices reduce the cognitive cost of making decisions. A world without prices—where you had to evaluate every good based on its intrinsic properties and your own needs—would be cognitively overwhelming. Prices provide a common metric for comparison.

As always, prices are never the *full *story. For example, brand names add another layer. Sowell uses the example of a Holiday Inn sign. It doesn’t promise the best hotel in town, just a predictable one. One to lower cognitive costs for the buyer. A traveler might find a better, cheaper independent motel if they searched long enough. But the brand reduces the variance of the outcome. You pay a premium to avoid inspecting five unknown motels. The brand has already done the sorting.

Alchian and Allen make the same point more precisely: brand names matter most for purchases that are “harder to evaluate before purchase” and where “subsequent surprising defects would cause more severe damage.” The brand reduces the shopper’s cost of estimating reliability.

These types of information-economizing institutions are everywhere. Culture works the same way. Sowell argues that culture provides “pre-packaged” beliefs that save the costs of reinventing the wheel on every decision. No one invents these heuristics from scratch. They circulate socially, get updated through use, and spread because people find them useful.

If cognitive constraints lead to systematic mistakes, what keeps those mistakes from overwhelming the whole system, leading to chaos? The answer is that reality eventually intrudes.

Start with the Alchian point about market competition. Firms that make consistently bad decisions, whether based on wrong beliefs about costs, demand, or competitors, tend to lose money and exit. This selection process doesn’t require any individual manager to be perfectly rational. It just requires that firms with better decision-making tend to survive longer. The market selects for behavior that approximates profit maximization, regardless of the actual decision processes inside firms. You don’t need to assume rationality to get rational-looking outcomes. You need selection pressure.

For individual consumers, the selection mechanism is weaker but still present. The budget constraint is real. You can use whatever mental accounting system you like, but you can’t spend more than you have, at least not for long. People whose mental accounting systematically exceeds their actual budget will eventually face consequences (overdrafts, debt, inability to pay rent) that force adjustment.

Wrong beliefs don’t get corrected immediately or completely. Competition works slowly and imperfectly. Some mistakes persist because the stakes are too low to justify the cognitive cost of fixing them. If your envelope budget is off by 20%, but the category only matters for $50, the expected loss is $10. Spending an hour doing precise calculations to save $10 isn’t worth it.

The heuristics that survive are the ones that work well enough in the domains where people actually use them. These heuristics persist because, for everyday discretionary spending, they produce acceptable results at low cognitive cost.

David Levine makes a related point in Is Behavioral Economics Doomed?. His complaint about behavioral economics is that it’s “obsessed with people being dysfunctional” and overlooks why behavior is functional—sometimes for subtle reasons, sometimes for obvious ones. “Seemingly dysfunctional behavior is often quite sensible when the circumstances and incentives are understood properly.” These are experienced consumers making familiar decisions in domains they’ve navigated for years. This isn’t a novel lab setting where we’d expect anomalies. It’s exactly the domain where standard economic analysis predicts behavior will look sensible.

Taking cognitive constraints seriously lets us go further. Not just “behavior is functional" but why and in which domains. Not everything is a familiar environment.

Bohren, Hascher, Imas, Ungeheuer, and Weber have a recent paper where they show how attention and memory constraints can generate actual mistakes. By mistake, they don’t mean failure to optimize some ideal function, but beliefs that are measurably wrong about objective probabilities. But you would only do this with some sort of cognitive constraint. The important part is that the constraint exists.

When information arrives all at once, attention is the binding constraint. You can’t process everything, so you focus on what stands out. Rare but dramatic events get overweighted. When information arrives over time, memory is the binding constraint. You can’t remember everything, so unusual occurrences fade. The same objective information produces opposite biases depending on how it’s presented. That’s a testable prediction: change how information arrives, change the direction of the error.

Here I circle back to: Institutions matter. In markets, the budget constraint provides a hard backstop. As I’ve written before, even Gode and Sunder’s “zero-intelligence traders” making random choices converged to a competitive equilibrium—the constraint did all the work. Becker made the same point about consumers: imagine people choosing randomly, subject only to their budget. When prices rise, the budget constraint rotates, forcing them to buy less of the expensive good on average. You get downward-sloping demand without any optimization. Whatever heuristics people use, they can’t systematically spend more than they have.

Reality bats last.

No posts