Partnership Across Dimensions

At the north end of the Bassin de la Villette in Paris stands a round stone building with a ring of columns, built in the 1780s by the architect Claude-Nicolas Ledoux. It looks like a small temple. It was a toll house, one of some sixty gates in a wall twenty-four kilometers long that the Ferme générale, the private company that collected the king’s indirect taxes, built around the city so that nothing could come in without paying the duty on it. Parisians said that le mur murant Paris rend Paris murmurant, the wall walling Paris makes Paris murmur. On the nights of July 12 and 13, 1789, before anyone marched on the Bastille, crowds burned most of the gates. The Rotonde survived, and today it’s a restaurant.

The wall had been proposed by one of the company’s own partners, Antoine Lavoisier, the chemist, who had bought a share in the Ferme in 1768 and used its income to fund his laboratory. Smuggling was eating into the company’s take, and a wall was the remedy. On May 8, 1794, Lavoisier went to the guillotine with twenty-seven other farmers-general, convicted of defrauding the people.

The Ferme générale was a public-private partnership in every sense the term carries today. The crown kept sovereignty, set the taxes, and owned the right to them. A syndicate of private financiers bid for a six-year lease, advanced the crown a large sum up front, and then collected what it could, keeping the difference. The king got money now and an administration he didn’t have to run. The financiers got a protected revenue stream and, in time, some of the largest fortunes in France. The people of Paris got a wall.

An industry for a form

The modern version goes by PPP or P3. A government signs a long contract, usually twenty to fifty years, under which a private consortium designs, builds, finances, operates, and maintains a road, a hospital, a school, a port, or a water system. The consortium is paid by user fees such as tolls, by “availability payments” from the government for keeping the asset open, or by a mix of both. What sets it apart from ordinary contracting is the bundle: one firm carries the project from drawing board through decades of operation, and takes on financing and some of the risk along the way.

There is a whole industry built around promoting this form. The World Bank runs a PPP resource center and keeps a database of private participation in infrastructure. Its International Finance Corporation advises governments on structuring deals. A multi-donor trust fund housed at the Bank, the Public-Private Infrastructure Advisory Facility, exists to rewrite countries’ laws and regulations so that such deals become possible, and has worked in more than a hundred and thirty of them. The Asian Development Bank has an office for it, the European Investment Bank hosts a European PPP Expertise Centre, the UN Economic Commission for Europe promotes what it calls “people-first PPPs,” and there is a World Association of PPP Units and Professionals in Geneva and a Canadian Council for Public-Private Partnerships with an annual conference. A practitioner can earn a credential, the Certified PPP Professional, developed with backing from the development banks. None of these bodies is selling a road. They’re selling the partnership itself.

Libertarians mostly don’t belong on that list, though they often get filed there. The Cato Institute’s position is that assets which can pay for themselves through user fees, such as airports, seaports, air traffic control, and many highways, should simply be privatized. Writers there have treated PPPs as a step partway in that direction, better than conventional public procurement when the private side bears real risk, and they’ve also published the critics, including Eduardo Engel, Ronald Fischer, and Alexander Galetovic, whose work shows that partnerships don’t free up public money the way their promoters claim. The partnership, in that view, is a way station on the road to the market.

That view assumes there is a road, one line with the state at one end and the market at the other, and that a partnership sits somewhere in the middle.

Not ends of one line

Public and private bodies don’t differ the way a hot day differs from a cold one, by degree along a single scale. They differ the way north differs from up. Each answers to its own set of pressures, and a move along one doesn’t register on the other.

A private firm in a competitive market is held to account by whoever holds its residual claim: the owners who keep what’s left after costs, and lose it when there isn’t anything left. That claim is concentrated. It shows up every quarter, it’s denominated in money, and it ends in bankruptcy if it goes far enough wrong. Customers can leave, rivals can undercut, lenders can call a loan. The feedback is sharp and narrow. It tells a firm whether people will pay for what it’s doing and almost nothing else.

A public agency answers to elections, auditors, courts, newspapers, unions, and whichever interest groups are paying attention that year. Its residual claim is spread across every future taxpayer and user, which means no one in particular holds it. An official who lets a bridge deteriorate rarely pays for it personally, and one who keeps it in good repair rarely gets credit. The agency is asked to serve many goals at once, from equity and access to visibility and legal compliance, and its feedback is broad, slow, and noisy. It tells an agency whether the public is angry and almost nothing about cost.

The differences run down a long list. What each side counts as success: a return on capital, or a set of qualitative goals no one ranks. How it decides: once capital is committed, or through layers of approval and veto. What it knows: engineering and finance in depth, or political constraints in depth. How hard its budget is: hard at every moment, or soft today and painful later. Whether it can leave: a firm can sell, exit, or walk away from a losing contract, while a government can’t abandon a hospital. None of these is a moral judgment of the people involved. They are the predictable behavior of people working under different selection pressures, and someone who moved from a city transit agency to a toll-road operator would start behaving differently within a year.

Two clocks

The sharpest of the mismatches is time.

A public asset lives a long time. A water main might be in the ground for eighty years, a bridge for a century. But the clock that governs public decisions runs on two- and four-year elections, annual budgets, and news cycles measured in weeks. So the public side tends to favor what can be opened with a ribbon before the next election and to underfund maintenance, whose cost is visible now and whose benefit arrives after the officials who paid for it are gone.

A private firm in a competitive market runs on a clock of its own, usually shorter than people expect. Its investors look for returns over three to seven years. Its managers are paid on this year’s results. It can’t assume it’ll still be the relevant player in twenty years, because a rival or a new technology may have replaced it, and so it demands a high return on any capital it has to lock up for a long time. The one condition under which a private firm can comfortably think in decades is monopoly: a regulated utility, a railroad with an exclusive right-of-way, a franchise no one else is allowed to hold. Monopoly, Natural and Otherwise traced how many such positions are handed out rather than won.

A long partnership has to bridge those clocks, and the only bridge available is to stretch the private side’s horizon to match the asset’s life. That means giving the private partner a monopoly for the length of the contract and protecting it against the public side’s own short clock. In December 2008 the Chicago City Council, given about two days to read the deal, leased the city’s 36,000 parking meters for seventy-five years to a consortium led by Morgan Stanley for about $1.15 billion. The city’s inspector general later estimated that the meters were worth nearly a billion dollars more than that. Rates rose sharply. The contract obliges the city to compensate the operator whenever it takes metered spaces out of service, for a street festival, a bike lane, a new bus stop, or a block closed for construction, so that every future change to how Chicago uses its own curbs now carries a bill payable to the lessee. The investors are reported to have recovered their outlay with more than half a century of the lease still to run. Chicago got its money in one year and gave up a say over its streets until 2083.

The Ferme générale ran on the same mismatch in the other direction. The crown’s claim to its taxes was perpetual, but the farmers’ leases ran six years, and a financier with six years to recover his advance had every reason to squeeze the most out of the people in that window and to let the long-run cost of their resentment fall on the king.

No halfway point

If public and private were two ends of one line, partial steps between them would make sense: move a little toward the market, then a little more. But the two don’t share a line, and so a partnership isn’t a point partway along one. It is something with a coordinate in each dimension at once, held there by contract. The contract tries to pull the private side’s time horizon and risk exposure out to where a public asset needs them, and to hold the public side to commitments its politics will want to break. Neither side stays put without enforcement, and the enforcement has to last as long as the contract does.

That is why lasting, successful partnerships are so rare. Large studies of infrastructure concessions in Latin America, led by J. Luis Guasch at the World Bank, found that most transport and water concessions were renegotiated, on average three and a half years after signing, and that the renegotiations usually favored the operator. Britain’s Private Finance Initiative built hundreds of hospitals and schools from the 1990s on, at unitary charges that often added up to several times the value of the buildings over the life of the contracts, before the government stopped signing new ones in 2018. That same year the contractor Carillion collapsed, and the new Royal Liverpool University Hospital, half built under a PFI deal, stood unfinished until the public side took it over and opened it in 2022. When a partnership comes under strain, the risk that was supposed to have been transferred goes back where it started, to the side that can’t walk away.

The conversions that did work look less like partnerships than separations. In 1996 Canada moved its air navigation system out of the Department of Transport into Nav Canada, a private nonprofit corporation that pays for itself through charges on the airlines that use it. Safety regulation stayed with Transport Canada. The operator lives entirely by the private side’s rules, with its own budget, its own capital, and its own users, and the regulator lives entirely by the public side’s, with no stake in the operator’s revenue. Neither is asked to behave like the other. Over the following decades Nav Canada modernized its equipment faster than the American system, which is still run by the Federal Aviation Administration, and handled growing traffic with relatively fewer staff. Delhi’s electricity distribution is a smaller case of the same thing: in 2002 the city sold majority stakes in its distribution companies under a separate regulator, and the share of power lost to theft and leakage, which had been above half, fell over the following years to a fraction of that.

These weren’t halfway steps. Each moved one bounded function all the way across into the private dimension and left a separate public body to watch it from the other. Bill of Separations argued that liberty is better kept by separating powers than by blending them. A partnership blends them by design, and a separation keeps each one whole in its own dimension.

Schools, hospitals, houses

Roads and air traffic control have something that makes separation easier: users who can be charged and output that can be counted. Schools, hospitals, and housing have neither in clean form, and here the partnership form does worst.

Britain’s PFI hospitals and schools carried the general pattern, faster construction in some cases, higher lifecycle costs in most, and contracts so rigid that a school wanting to move an electrical outlet could find itself paying the operator’s rate to do it. In the United States, studies of public hospitals converted to private ownership have found that profitability rose mostly through fewer staff, higher prices, and a smaller share of poorer patients, and a 2024 review in The Lancet Public Health of evidence from high-income countries found that outsourcing and privatization of health care were associated with worse outcomes in most of the studies it covered. The private partner did what a private residual claim rewards, and none of it was what the public side had wanted.

What worked better in these fields wasn’t a long partnership with a provider but money that follows a person. A housing voucher pays part of a family’s rent in whatever private apartment it finds, and studies have generally found it delivers more to the family per public dollar than public housing does. An education voucher or savings account lets a student’s share of school funding go to a school the family chooses. The results on test scores have been mixed, better in small, targeted programs than in large recent ones, and parents’ satisfaction has been consistently high. Either way, there’s no thirty-year contract binding a ministry to a firm. The public side raises and pays the money in its own dimension, the private provider competes for the family in its own, and the family carries the money across. The separation holds because the one party who moves between the two dimensions is the person the service is for.

The substrate

The Ferme générale’s real product was never tax collection. It was the space between the crown and the company, where neither the king’s officials nor any market could see clearly what was owed to whom.

A modern partnership opens the same space. Its contract runs for decades and can’t specify everything that will happen, so it has to be renegotiated, and each renegotiation is a private deal over public money. Once construction has begun, cancelling is politically and fiscally painful for the government, which hands the operator a strong hand at every reopening. A consortium can win the award with a low bid and recover the difference later through change orders, revised demand guarantees, or new termination terms, and the gap between the original bid and the renegotiated outcome is a rent in the plainest economic sense. The private side usually knows more about costs, while the public side controls land, permits, and the threat of non-payment, so either can hold the other up, and a side payment is often cheaper than a fight.

The accounting helps. Availability payments and revenue guarantees keep the obligation off the government’s main balance sheet, so the official who signs the deal gets the ribbon-cutting and a successor inherits the bill. Every stage of the bundle, from land assembly to financing, construction, operation, refinancing, and hand-back, is a new negotiation with lopsided information. Studies of developing countries have found that more PPP deals get signed where measured corruption is higher, which is what one would expect if the form attracts the people best placed to profit from its gaps.

None of this requires anyone to be a villain. In a public agency the residual claim is diffuse but someone, at least nominally, answers for it at the next election. In a private firm the owners answer for their losses with their own capital. In a partnership the residual is split between them and blurred, a contractual claim on one side, a political liability on the other, and the losses spread across taxpayers, users, and governments not yet elected. Value can move from one side to the other through that blur without passing through a market’s discipline or a vote. The ice trust in Monopoly, Natural and Otherwise had the mayor of New York and his brother among its shareholders, and it held the city’s ice supply through landing slips the city’s own Dock Department had granted it.

Ways across

If a partnership holds an activity between the two dimensions, the transitions that have worked are the ones that carried it all the way into the private one, a bounded piece at a time, and left nothing behind to renegotiate.

Japan’s national railway is the largest case. By 1987 Japanese National Railways carried debts of more than thirty-seven trillion yen and lost money on almost every line outside the Tokyo–Osaka corridor. The government didn’t bring in a private partner to run it under a long contract. It broke the railway into six regional passenger companies and a freight company, moved the old debt and surplus land into a separate public settlement corporation, and let the new companies start clean. The three companies on Honshu were sold to private shareholders over the following two decades, JR East by 2002, JR West by 2004, and JR Central by 2006, and JR Kyushu followed in 2016. The liabilities of the old public railway stayed in the public dimension, where they were wound down, and the operation went into the private one with its own revenue, its own capital, and no contract binding it to a ministry. The companies on the smaller islands, Hokkaido and Shikoku, whose lines can’t pay for themselves, are still owned by the state, and nobody pretends otherwise.

A second kind of transition removes the public monopoly before it does anything else. Britain’s Transport Act of 1980 ended the licensing that had protected the state-owned National Bus Company on long-distance coach routes. No asset was sold and no contract was signed. Private operators were simply allowed to run coaches between cities, and within months fares on the main routes had fallen sharply while the state’s own coaches, facing rivals for the first time, cut theirs to match. Entry did what a partnership’s performance clauses try and mostly fail to do. Transitioning to Private Pickleball found the same thing at a much smaller scale: private clubs appearing beside public courts, competing on what the public courts couldn’t offer, without any city having to hand anything over.

Where the activity sits on a chokepoint, the land and the operation can go to different dimensions. A railway right-of-way, a port’s waterfront, or a river crossing is a location monopoly, and selling it outright hands a private owner a toll on everyone who has to pass. Leaving it with an agency that also runs the trains or the ships brings back every problem of public operation. The Georgist separation splits the two. The community keeps the land and collects its rent, which Monopoly, Natural and Otherwise argued belongs to the community whose presence created it, and private operators own and run what stands on it, under leases short enough that the land can be let again to someone else when they stop serving it well. The rent flows one way, the operation answers to its own customers, and there’s no thirty-year bundle where the two are traded against each other.

Where a function stays public in purpose, the clean transitions keep any contract short and the public side’s exit real. Since 1985 London has put its bus routes out to tender, route by route, for terms of about five years. Transport for London sets the routes, the timetables, and the fares, and collects the fares itself. Private companies own the buses, hire the drivers, and bid to run each route, and a company that runs one badly loses it at the next tender. That’s ordinary contracting, a payment for a defined service over a period that matches a competitive firm’s own clock, and it asks neither side to take on the other’s nature. The voucher works the same way from the other direction, with the renewal decided not by an agency at a tender but by each family every year.

The wall

Lavoisier wasn’t a thief by the standards of his day. He kept careful books, argued for reforms in how the Ferme treated the people it taxed, and spent his share of the profits on chemistry. His wall was a sound answer to the company’s problem, which was smuggling, and a terrible answer to the city’s. The Revolution abolished the Ferme générale in 1791 and its partners’ heads followed three years later, but the city kept collecting the entry duty, the octroi, until 1943. Several of Ledoux’s toll houses still stand, at Denfert-Rochereau, at the Place de la Nation, in the Parc Monceau, and at La Villette, where people now sit outside the Rotonde in the evening with a drink, on the line where every cart entering Paris once stopped to pay.

Sources

• Eugene N. White, “From Privatized to Government-Administered Tax Collection: Tax Farming in Eighteenth-Century France,” Economic History Review 57, no. 4 (2004).
• Anthony Vidler, Claude-Nicolas Ledoux: Architecture and Social Reform at the End of the Ancien Régime (1990).
• Arthur Donovan, Antoine Lavoisier: Science, Administration, and Revolution (1993).
• Eduardo Engel, Ronald Fischer, and Alexander Galetovic, The Economics of Public-Private Partnerships: A Basic Guide (2014).
• J. Luis Guasch, Granting and Renegotiating Infrastructure Concessions: Doing It Right (World Bank, 2004).
• Chris Edwards, Cato Institute, writings on infrastructure privatization and PPPs (Downsizing the Federal Government).
• World Bank PPP Resource Center; Private Participation in Infrastructure (PPI) database; Public-Private Infrastructure Advisory Facility (PPIAF); European PPP Expertise Centre (EPEC); UNECE, “People-first PPPs.”
• City of Chicago Office of Inspector General, Report on the Chicago Parking Meter Concession (June 2009).
• UK National Audit Office, PFI and PF2 (January 2018); HM Treasury, Budget 2018, on ending new PFI contracts.
• Civil Air Navigation Services Commercialization Act, S.C. 1996, c. 20, establishing Nav Canada.
• Benjamin Goodair and Aaron Reeves, “The Effect of Health-Care Privatisation on the Quality of Care,” The Lancet Public Health 9, no. 3 (2024).
• Fumitoshi Mizutani and Kiyoshi Nakamura, “The Japanese Experience with Railway Restructuring,” in Governance, Regulation, and Privatization in the Asia-Pacific Region (NBER, 2004), on the 1987 breakup of Japanese National Railways.
• Transport Act 1980 (UK), on deregulation of express coach services; Transport for London, London’s Bus Contracting and Tendering Process.
• Elisabetta Iossa and David Martimort, “Corruption in PPPs, Incentives and Contract Incompleteness,” International Journal of Industrial Organization (2016).

---

A note on how this piece was written: the subject, the questions it asks, and the positions it takes are mine, worked out in a research conversation with an AI model. Much of the research, the examples, and the sentences were drafted by an AI model working from that direction, and then edited by hand. I’d rather say that plainly than have a reader guess at it.