There is a particular kind of economist who looks at the world and sees not just markets, but mechanisms — systems that can be tuned, adjusted, calibrated. These are the engineering marginalists, and they are perhaps the most consequential thinkers you've never heard of. They don't just want to understand the margin. They want to build things on it.
The story of engineering marginalism begins, fittingly, with an actual engineer standing on an actual bridge, staring down at the Seine.
Dupuit's Bridge: The Origin of Engineering Marginalism
A French engineer discovers that pricing at the margin, not average cost, determines efficient allocation.
Jules Dupuit was not an economist by training or temperament. He was a French civil engineer, a builder of roads and bridges and aqueducts, a man who spent his career in the Corps des Ponts et Chaussées — the Corps of Bridges and Roads, which is exactly the sort of institution that could only exist in France. [C1] In the 1840s, Dupuit found himself wrestling with a question that sounds mundane but turns out to be profound: How much should you charge people to cross a bridge?
The conventional wisdom was simple. You built a bridge. You added up what it cost. You divided by the number of people who would cross it, and you charged each of them that amount. Cost recovery. Clean, fair, obvious.
Dupuit thought this was idiotic.
His insight was that the value of crossing a bridge is not the same for everyone. A merchant hauling goods to market might value the crossing enormously — it saves him days of travel on an alternate route. A local farmer visiting his cousin might value it only slightly. If you set the toll high enough to recover costs, you exclude the farmer entirely, even though letting him cross costs essentially nothing. The bridge is already built. The marginal cost of one more pedestrian walking across it is, for all practical purposes, zero.
This is the key move: Dupuit separated the cost of building the bridge from the cost of using the bridge. And he realized that efficiency demands you price at the margin — at the cost of one additional crossing — not at the average. The fixed costs of construction are, as economists say, sunk. Gone. Irrelevant to the pricing decision. What matters is the marginal cost, and for a bridge that's already standing, the marginal cost of one more crosser is close to nothing.
From this observation, Dupuit developed what we now recognize as an early version of consumer surplus — the idea that people are often willing to pay more than they actually have to, and that the gap between willingness to pay and actual price represents a kind of invisible wealth. [C2] He was doing this in 1844, a full three decades before the so-called Marginal Revolution supposedly began. The economists were, as usual, late to their own party. The engineer had gotten there first.
Dupuit went on to apply marginal analysis to the Paris water supply and sewer system, arguing for variable pricing based on use rather than flat fees. He was, in the fullest sense, an econo-engineer — someone who understood that the abstract beauty of marginal thinking was only as good as its ability to improve actual systems in the actual world. He wanted better bridges, better sewers, better roads. Marginalism was his tool, not his religion.
This engineering sensibility — the conviction that marginal analysis should be applied, not merely admired — runs like a deep current through the history of economics, surfacing in unexpected places and producing some of the most important policy innovations of the modern world.
Stockholm's Experiment: Theory Meets Practice
A congestion pricing trial transforms public opinion through lived experience of marginal incentives.
Let me take you to Stockholm in January 2006.
The Swedish capital had a traffic problem. Not an unusual one — the same creeping, thickening, soul-deadening congestion that afflicts every prosperous city that made the mistake of building roads in the twentieth century. [C3] Cars poured into the city center each morning and sat there, idling, emitting, making everyone miserable in that particular way that traffic makes you miserable: slowly, invisibly, one wasted minute at a time.
The city's politicians proposed congestion pricing — charging drivers a fee to enter the central zone during peak hours, with the price varying by time of day. Higher during the morning and evening rush, lower during midday, free at night. Pure Dupuit, applied to asphalt instead of stone.
The public hated the idea. Polls showed clear opposition. People were viscerally offended by the notion of paying for something they'd always done for free. Never mind that they were already paying — in time, in stress, in the particulate matter settling into their children's lungs. Those costs were invisible, diffuse, marginal in the worst sense: too small in any given moment to trigger outrage, but enormous in aggregate. The proposed toll, by contrast, was concrete, specific, and felt like a shakedown.
The city did something remarkable. They ran the trial anyway.
For seven months, from January to July 2006, Stockholm operated its congestion pricing system as an experiment. Cameras photographed license plates at entry points. Bills arrived in the mail. The maximum daily charge was about 60 Swedish kronor — roughly six dollars at the time, less than the price of a decent sandwich. [C4]
The results were immediate and dramatic. Traffic into the central city dropped by 22 percent. Not because people stopped going to work or shopping or living their lives, but because the marginal incentive pushed them to adjust at the margin. Some commuters shifted to public transit. Some shifted their travel times by twenty minutes, slipping in just before or after the peak pricing window. Some carpooled. Some combined errands. Each individual adjustment was small — marginal, you might say — but the aggregate effect was transformative. Air quality improved measurably. Commute times fell. The city felt, according to residents, simply better.
And then, in September 2006, the experiment ended. The cameras went dark. The tolls stopped.
Traffic roared back immediately. Within weeks, congestion had returned to pre-trial levels. The city held a binding referendum.
Here is the beautiful part: the same citizens who had opposed congestion pricing in polls before the trial now voted to make it permanent. They had lived inside the system. They had felt the difference. The abstract argument — that pricing negative externalities at the margin improves everyone's welfare — had failed to persuade them. The lived experience of marginalism in action had succeeded.
Stockholm's congestion pricing system has been running ever since, and the city's experience has become one of the most cited case studies in urban economics. [C5] It demonstrates something crucial about engineering marginalism: it often works better than people expect, and people often can't know that until they see it. The benefits of marginal pricing are emergent. They arise from thousands of small adjustments that no single person can observe or predict. You can't intuit your way to the answer. You have to build the system and let it run.
Singapore's Systematic Approach: Engineering the City
Dynamic pricing of roads, ownership, and even fines transforms Singapore into a laboratory of applied marginalism.
Singapore understood this decades before Stockholm, and they understood it more completely than anyone.
The island city-state introduced congestion pricing in 1975 — Area Licensing Scheme, they called it, with the bureaucratic elegance that characterizes Singaporean governance. [C6] It started as a simple system: buy a paper license to drive into the central business district during restricted hours. Over the years, it evolved into something far more sophisticated. The current system, called Electronic Road Pricing, uses overhead gantries and in-vehicle transponders to charge variable rates that change every half hour based on real-time traffic conditions. The price of driving into the city at 8:30 a.m. on a Monday is different from the price at 9:00 a.m., which is different from 9:30 a.m. It is dynamic marginal pricing in its purest form, and it works spectacularly well.
But Singapore's engineering marginalism extends far beyond road pricing. The country imposes enormous taxes on car ownership itself — a Certificate of Entitlement that can cost more than the car, registration fees, import duties, and various other levies that together can make a Toyota Corolla cost upwards of $100,000. [C7] This sounds insane until you remember that Singapore is a tiny island with limited road space, and that every additional car imposes real costs on every other person. The taxes are, in principle, an attempt to make private car owners internalize the full social cost of their choice.
And then there are the fines. Singapore is famous for its fines, and rightly so. But consider what a fine actually is from a marginalist perspective. There is a sign in Singapore that reads, roughly, "$250 plus GST" — Goods and Services Tax — for some particular infraction. Think about what that "plus GST" means. It means the government is treating the fine not as a punishment but as a price. It's a consumption tax on illegal activity. You are purchasing the right to break the rule, and the government is taxing your purchase. This is engineering marginalism taken to its logical, and perhaps slightly unnerving, extreme. It treats everything — even transgression — as a transaction with a marginal cost that can be optimized. [C8]
Vickrey's Mechanism: Building Systems for Truthfulness
Elegant mechanism design makes honesty the selfish choice in auctions and resource allocation.
The spirit of Dupuit — the engineer who prices things correctly — finds its most elegant modern expression in the work of William Vickrey. Vickrey was a Canadian-born economist at Columbia who spent his career thinking about mechanism design: how to build systems where people's selfish incentives align with socially desirable outcomes. He won the Nobel Prize in 1996, three days before he died, which is either tragic or poetically efficient depending on your disposition. [C9]
Vickrey's most famous contribution is the second-price auction, now called the Vickrey auction. The rules are simple: everyone submits a sealed bid. The highest bidder wins but pays the second-highest bid. This sounds like a quirky technicality, but it's actually a masterpiece of marginal incentive design.
In a standard sealed-bid auction, you face an agonizing strategic problem. If you bid your true value, you might win but pay too much. If you shade your bid down, you might lose an item you'd have been happy to pay for. Your optimal bid depends on what you think everyone else will bid, which depends on what they think you'll bid, and so on into an infinite regress of strategic calculation.
The second-price auction dissolves this problem entirely. Your dominant strategy — the best thing to do regardless of what anyone else does — is simply to bid exactly what the item is worth to you. If you win, you'll pay less than your bid (you pay the second-highest price, remember), so you always get a surplus. If you lose, it's because someone valued it more than you did, and you wouldn't have wanted to pay their price anyway. The mechanism makes honesty not just virtuous but optimal. Truth-telling becomes the selfish choice. [C10]
This is engineering marginalism at its finest. You don't need to change human nature. You don't need to lecture people about fairness or social welfare. You build a mechanism where the marginal incentives point toward the outcome you want, and then you let self-interested people do what self-interested people do. The system does the moral work.
Google, it's worth noting, used a modified Vickrey auction for years to sell its search advertising — the most valuable real estate in the history of commerce, priced by a mechanism invented by a mild-mannered Columbia professor who was thinking about stamp collectors and spectrum licenses. Dupuit would have been delighted.
The Invisible Hand at Work: Emergent Incentives
When multiple programs interact, marginal incentives emerge—often with unintended consequences.
Once you start seeing the world through the lens of engineering marginalism, you find applications everywhere, and some of them are uncomfortable.
Consider Social Security's retirement earnings test. For years in the United States — and to some degree still — retirees who continued working past a certain age and earning above a certain threshold would lose Social Security benefits at a rate of roughly 50 cents for every additional dollar earned. Combined with income taxes and payroll taxes, an elderly worker could face an effective marginal tax rate of 80 percent or more on additional earnings. [C11] Earn an extra dollar, keep twenty cents. The incentive structure was screaming at older Americans to stop working, stop contributing, stop being productive. And many of them listened, because people respond to marginal incentives, even when — especially when — nobody explicitly explains the incentives to them.
This wasn't anyone's deliberate policy. Nobody sat in a committee room and said, "Let's design a system that discourages the elderly from working." It emerged from the interaction of multiple well-intentioned programs, each sensible in isolation, that together created a perverse marginal incentive. This is one of the most important lessons of engineering marginalism: systems have emergent properties. The marginal incentives that actually govern behavior may be invisible to the people who designed the system and the people who live within it.
The engineering marginalist's response is not to moralize but to redesign. Change the formula. Adjust the phase-out rate. Shift the kink point. These sound like technocratic details, and they are technocratic details, and technocratic details are where an enormous share of human welfare is actually determined. The tax code is full of implicit marginal incentives that most citizens never see and most legislators barely understand, and those invisible incentives shape behavior as surely as any law or sermon. [C12]
Pricing Externalities: From Carbon to Congestion
Engineering marginalists close the gap between private and social costs through taxes and fees on spillover effects.
The boldest applications of engineering marginalism are in the pricing of externalities — those costs and benefits that spill over onto third parties who never agreed to the transaction.
A carbon tax is the purest example. You burn fossil fuel, and the carbon dioxide enters the atmosphere, warming the planet, raising sea levels, shifting weather patterns, imposing costs on billions of people who had no say in your decision to drive to the store for milk. The marginal private cost of your gallon of gasoline — what you pay at the pump — is far below the marginal social cost, which includes the climate damage. A carbon tax closes that gap. It makes you pay, at the margin, for the harm your consumption imposes on others.
The logic extends to practically anything with spillover effects. Noise taxes — making the person playing music at 2 a.m. pay for the sleep they're destroying. Obesity-related taxes on sugary drinks — controversial, certainly, but rooted in the observation that the marginal cost of a soda doesn't include the downstream healthcare costs that are partly socialized through insurance. [C13] Congestion pricing, which we've already discussed, is really just an externality tax applied to the specific externality of taking up road space.
Each of these proposals follows the same template: identify a gap between private marginal cost and social marginal cost, then design a mechanism — a tax, a fee, a pricing scheme — that closes the gap. The engineering marginalist doesn't ask whether people are good or bad, greedy or generous. She asks whether the prices are right. And when they're not, she fixes them.
There is something both thrilling and unsettling about this worldview. Thrilling because it suggests that many of our worst collective problems — pollution, congestion, resource depletion — are not failures of human character but failures of pricing. We don't need better people; we need better prices. Unsettling because it implies a world in which everything is, in principle, priceable, and the highest calling of the economist is to be a kind of universal meter-reader, attaching the correct marginal cost to every human activity.
Not everyone is temperamentally suited to this vision. Among economists, there's a spectrum. Some are drawn to grand theory, to the elegance of general equilibrium, to the contemplation of how markets coordinate information across millions of agents. Others are drawn to the engineering side — to the hands-on work of designing auctions, calibrating taxes, building mechanisms. My co-blogger Alex Tabarrok is firmly in the latter camp, more of an engineering marginalist than I am. Alex would happily spend a week designing the optimal congestion pricing schedule for the Washington Beltway. I would more likely spend that week reading about a seventeenth-century philosopher who anticipated the concept while discussing the morality of bridge tolls. Both approaches have their place, but it's the engineers who change the world. The rest of us just explain why they were right. [C14]
From Theory to Practice: Engineering's Gift to Economics
The convergence of engineering and marginalism created a proving ground for economic theory.
The late nineteenth century saw a remarkable convergence between engineering and economic thinking. The great marginalists — Jevons, Walras, Marshall — were working in an era of extraordinary engineering achievement. Bridges, railroads, telegraph networks, municipal water systems. The physical infrastructure of modern life was being built at a pace never before seen, and the people building it needed to make decisions about pricing, allocation, and investment. They needed, whether they knew it or not, marginal analysis.
It is no coincidence that marginalism emerged in this context. When you are building a railroad, you need to know the marginal cost of carrying one more ton of freight. When you are operating a telegraph network, you need to know the marginal cost of transmitting one more message. When you are supplying water to a city, you need to know the marginal cost of one more gallon. These are not abstract questions. They are practical, urgent, and consequential. Get the marginal cost wrong and you bankrupt the railroad or leave the city thirsty.
The engineering mindset — build something, measure the results, adjust — gave marginalism a proving ground. Dupuit's bridges, Vickrey's auctions, Singapore's road pricing, Stockholm's congestion experiment: in each case, the theory was validated not by its internal elegance but by its performance in the field. Engineering marginalism is, at its core, an empirical enterprise. It says: here is a system, here are its marginal incentives, here is what happens when we change them. The proof is in the outcome.
This is perhaps why Stockholm's congestion pricing story resonates so deeply. The citizens didn't need to understand consumer surplus or Pigouvian taxes or deadweight loss. They didn't need to read Dupuit or Vickrey. They needed to live for seven months in a city where the marginal cost of driving had been corrected, and to feel the difference in their commutes, in the air, in the texture of daily life. The theory was invisible. The results were not.
And that, in the end, is the promise of engineering marginalism: not that it will make the world theoretically optimal, but that it will make the world observably better, one corrected margin at a time. [C15]