President Trump has reportedly expressed reservations about public-private partnerships, but White House economic advisor Gary Cohn is still enthusiastic about building the administration's fabled infrastructure plan around them. Not everyone realizes, however, that there are two very distinct kinds of public-private partnerships, which I call the good kind and the bad kind. I'd like to believe that it is the bad kind that worries Trump while it is the good kind that encourages Cohn.
The good kind of public-private partnership is more formally known as a demand risk partnership. In this case, the public partner essentially gives the private partner a franchise to build a road or some other infrastructure. The private partner is allowed to collect tolls or other revenues from the infrastructure for a fixed period of time, usually three or four decades, after which ownership and management of the infrastructure is turned over to the public partner (who may contract it out again). The key is that private partner accepts all of the risk that the revenues may not cover the costs. The I-495 Capital Beltway express lanes are a demand risk partnership.
The bad kind of public-private partnership is more formally known as an availability payment partnership. As with the good kind, the public partner designs the project and the private partner builds and, usually, operates it. Unlike the good kind, the private partner takes no risk that the project might not pay its way. Instead, the public partner contracts to pay the private partner enough money over several decades to completely repay the private partner's costs regardless of whether anyone is actually using the infrastructure.
Availability payment partnerships might make sense in the case of infrastructure that no one expects to earn user fees, such as common schools. But in most cases, such partnerships are formed mainly to allow the public partner to sidestep legal debt limits. For example, euro nations are supposed to limit their debts to a fixed percentage of GDP. Some nations, such as Italy, have built high-speed rail and other infrastructure using availability payment partnerships so that the debts appear on the books of the private partners, not the government.
For the same reason, Denver's Regional Transit District (RTD) formed a public-private partnership to build a billion-dollar rail line to the airport. Voters had approved a sales tax increase for the rail line but set a debt limit. When cost overruns made it impossible to build the line without exceeding the debt limit, RTD entered into an availability payment partnership so the debt wouldn't appear on its books. Of course, it was still contracturally obligated to pay the private partner enough to repay its debt.
Demand risk partnerships are good because the need to cover costs out of user fees creates a discipline that insures that projects are worthwhile and costs do not get out of control. User-fee funded projects are also better maintained because managers know users will stop paying if the project becomes dangerous or unreliable.
Availability payment partnerships are bad because they offer little incentive to insure that projects are worthwhile or to control costs. Denver's airport rail line was originally projected to cost $315 million and ended up costing over a billion dollars. Nor did that billion dollars buy a quality product: more than a year after it opened, the builder still has not got the automatic grade crossing gates working, a technological problem that the private railroads solved more than 80 years ago.
Until the Trump administration releases its infrastructure plan, we won't know if it will make a distinction between demand risk and availability payment partnerships. But it should favor the former over the latter to protect taxpayers and insure that the infrastructure we build is both worth having and well maintained.
The wreck of the 501--the Amtrak train that crashed near Seattle on Monday--is raising lots of questions about Amtrak operations, but they aren't always the right ones. Here are some questions that should be asked and some of my preliminary answers. Answers from Amtrak (the operator), FRA (the funder), Sound Transit (the track owner), or WSDOT (the train owner) may differ.
1. Congress required passenger railroads to install positive train control (PTC) by the end of 2015. Why did the Federal Railroad Administration (FRA) give money to the Washington Department of Transportation (WSDOT) for a new passenger rail line that would not open until after 2015 when the project didn't guarantee funding for positive train control?
Answer: The Obama administration wanted to distribute high-speed rail funds to as many states as possible in order to build political backing for the program, so it couldn't be bothered with positive train control. The tracks the train was on are owned by Sound Transit, which says it is installing PTC, but it won't be finished until spring. Public releases of WSDOT's application for funds for this train didn't mention PTC.
2. Around 800 people die in railroad accidents a year. PTC would prevent only about 1 percent of these fatalities; far more would be saved by spending the same amount of money on better grade crossings and fencing of rail rights of way. Why do we put so much emphasis on an expensive technology that will do so little?
Answer: Accidents that PTC could have prevented tend to be more spectacular than people getting killed when a train hits their car at a grade crossing. This suggests that, when politicians decide where private businesses spend their money, it'll get spent on grandiose programs rather than things that could really make a difference.
3. When an auto driver runs a red light and kills a pedestrian, we don't blame the auto maker for not making driverless cars sooner; we blame the driver and, perhaps, the people who were supposed to train the driver. Why blame this accident on the lack of positive train control when the train driver should have slowed down and Amtrak should have made sure the driver was qualified to operate this section of track?
Answer: Everyone is looking for a scapegoat, and it is easier to blame an institution than an individual.
4. The train in question had about 250 seats, and this was the inaugural run on this route, which usually generates a lot of interest among rail enthusiasts. Yet there were only 80 passengers on board. Does this confirm that Lakewood Mayor Don Anderson was right when he said "this project was never needed"?
Answer: The big change in Seattle-Portland service was not an increase in speeds but an increase in frequencies (a change that may be delayed by the wreck of one of the new train sets funded by the federal government). Amtrak has only been able to fill 54 percent of this train's seats, and WSDOT was hoping that more frequent and more reliable trains would increase the percentage of seats filled. We'll know more after a year or so, but it doesn't look good if the inaugural run filled less than a third of the seats.
5. Why do so many reporters call this a high-speed train? The top speed between Portland and Seattle is 79 mph, the same as it has always been and the same as most other Amtrak routes. In technical terms, this was a conventional, low-speed train.
Answer: Though this was a low-speed train, it was funded by Obama's high-speed rail fund. By repeatedly using the term "high-speed trains," reporters are keeping that idea in the public consciousness, perhaps in the hopes that Trump's infrastructure plan will include money for more such trains. (This could backfire, however, by making people think that high-speed trains are more dangerous. They aren't--but they are a lot more expensive.)
6. Why do we need passenger trains at all? Amtrak fares from Seattle to Portland start at $26 and cover less than half the costs of the train. Bolt Bus has six buses a day that take less time than the train at fares of around $15. Plane fares start at $65, though most are around $100 (which may still be less than the full costs of Amtrak), and there are dozens of flights a day.
Short-distance trains were made obsolete by buses in the 1920s. Long-distance trains were made obsolete by planes in the 1950s. When other transportation technologies, such as horseback riding, steamboats, and canals went obsolete, we let them go. Why can't we let go of the passenger train?
Answer: America was suffering an inferiority complex in the early 1960s. We were losing the space race; some thought there was a missile gap with Russia; Japanese electronics were beginning to take over American markets. When Japan introduced its bullet trains in 1964, suddenly there was one more area in which our technology appeared to be inferior. Never mind that our jet airplanes were several times faster than Japan's trains; Congress began funding passenger trains in 1965, and once a federal program gets started, it generates special interest groups dedicated to keeping it going.
Question: At least Amtrak is getting closer to covering its operating costs, right?
Answer: No, when Amtrak says that, it is lying. Amtrak counts more than $200 million in annual subsidies that it gets from the states as "passenger revenues." Amtrak also pretends depreciation is zero even though, at more than $800 million per year, it is the second largest line item on its operating budget. Amtrak's deferred maintenance has led to a backlog of needs in the tens of billions of dollars. When counting only ticket fares and on-board food service revenues against operating costs, including depreciation, Amtrak operations lose more than $1 billion a year and ticket fares cover only a little more than half the costs.
Question: At least Amtrak's Northeast Corridor makes money, right?
Answer: Only if you don't count depreciation, deferred maintenance, or other costs that Amtrak doesn't try to allocate to individual routes. The Northeast Corridor needs at least $35 billion in rehabilitation work just to bring it up to a state of good repair. Another way Amtrak has made its trains appear to be profitable is by calling much of its maintenance work a capital cost, and Amtrak can't afford to do all of the "basic infrastructure" maintenance needed in the Northeast Corridor, much less the rehabilitation work, without federal subsidies of around $500 million dollars per year (see page 15).
Question: So is Amtrak's maintenance backlog only in the Northeast Corridor?
Answer: No, Amtrak doesn't own most of the tracks it uses outside of the Northeast Corridor, but it still has maintenance needs for its stations and rolling stock. When Amtrak took over private passenger service in 1971, the average age of passenger cars that it acquired from the railroads was 22 years, and they were so worn out that nearly all were replaced within a decade. Today, the average age of Amtrak's passenger cars is more than 30 years, suggesting that it will soon need to spend billions replacing them.
Question: So are all trains obsolete?
Answer: No, only passenger trains are obsolete. Freight trains are extremely productive, and America has the finest, most advanced rail system in the world. That's because it is mostly private and operates to produce profits, not to give politicians ribbon-cutting opportunities.
Today is not a proud day for the Washington State Department of Transportation (WSDOT). The agency spent close to $800 million of federal funds on a so-called high-speed rail project between Seattle and Portland--only "so-called" because top speeds would be just 79 mph, which is conventional rail. Much of the money was spent upgrading existing tracks to give passenger trains a shorter (but less scenic) route through and around Tacoma.
As you probably know, the very first train to use this route derailed on an overpass over Interstate 5, blocking half the freeway and killing at least six people. To make matters worse, Mayor Don Anderson of Lakewood, Washington, about 10 miles north of the crash, warned WSDOT a few weeks ago that it was not taking safety seriously enough. "This project was never needed and endangers our citizens," he cautioned.
To be fair, Mayor Anderson was worried that grade crossings in Lakewood were inadequately protected for 79-mph trains. But his comments more generally suggest that WSDOT was putting the goal of saving Seattle-Portland passengers ten minutes of time--increasing average speeds by just 2.7 mph--ahead of safety.
In response to the accident, President Trump tweeted, "The train accident that just occurred in DuPont, WA shows more than ever why our soon to be submitted infrastructure plan must be approved quickly." The implication was that this is an example of crumbling infrastructure, when in fact it is an example of misplaced infrastructure priorities.
In fact, what the accident shows is why the federal government should get out of the infrastructure business. As Mayor Anderson said, this project was unnecessary, and it was only done because President Obama wanted to spend billions of federal dollars on ideologically driven high-speed rail projects and WSDOT had a shovel-ready project (despite not being high-speed rail) on which to spend some of those dollars.
Trump's inclination is to have the federal government back out of the infrastructure-funding game. But many members of Congress in both parties see infrastructure as a store full of candy they can give out to please their constituents and campaign contributors. The danger is that a hastily passed bill will end up spending more billions on unnecessary projects like the Seattle-Portland train.
No matter what speed, intercity passenger trains are obsolete and have been at least since the advent of jet airliner service. Even after hundreds of millions spent on improvements, this particular train would have been slower than driving from Seattle to Portland, but even the fastest high-speed trains are slower than flying.
One airline alone offers nearly two dozen flights a day between the Portland and Seattle airports. People who complain that Sea-Tac Airport is a long way from Seattle would do better to seek more flights out of King County Airport (aka Boeing Field), which is just four miles from downtown Seattle, than to support more trains.
Amtrak's Seattle-Portland fare is $26 while the cheapest flights are $65. But Amtrak is heavily subsidized by both federal and state governments. Amtrak's Seattle-Portland trains (which also go to Vancouver, BC and Eugene, OR) earned just under $30 million in ticket revenues in 2016 but cost Amtrak more than $68 million to operate not counting depreciation and other costs that Amtrak doesn't allocate to individual trains. Meanwhile, subsidies to airlines are small and mostly go to support small-town airports.
On average and including all subsidies, airline travel costs about 16 cents per passenger mile while Amtrak costs about 60 cents per passenger mile. Higher-speed trains may attract more passengers but cost so much more that the costs per passenger mile are at least six times as much as the costs of flying.
Regardless of what you think of Amtrak, the point is that transportation spending decisions should be made by and in response to transportation users, not by politicians, and especially not by federal politicians. The accident in Dupont, Washington was a horrible tragedy, and we don't yet know exactly what caused it. But, no matter what the cause, it never would have happened were it not for federal involvement in infrastructure spending.
The Trump administration’s proposal to repair and expand America’s roads, bridges, ports and airports includes the expanded use of public-private partnerships (P3s). Under P3s, state and local governments award franchises to private companies that agree to pay for and manage the infrastructure in exchange for the companies receiving toll payments from future users. A number of P3 projects currently operate in the United States, and they are common in other developed nations.
Despite the growing embrace of these projects by policymakers around the world, the Trump proposal is being met with skepticism. For example, the New York Times dropped this article last week ahead of Trump administration efforts to promote the proposal. According to the article, “experts agree” that “there is little hard evidence” that such projects produce long-term benefits to the public as compared to traditional government-provided infrastructure. (That “agreement” came as news to many transportation experts.)
At heart, the article charges that P3 programs are “win/no lose” proposals for the private firms: if the projects prove popular, the firms profit—sometimes handsomely, to the detriment of consumers. But if the new infrastructure doesn’t get many toll-paying users, the financial losses from the projects fall on taxpayers.
To illustrate this, the NYT cites California State Route 91, one of the first P3s in the United States. Initially intended to reduce congestion, the project awarded a private company the right to build and operate a special four-lane toll road in the middle of the highway. The road was “congestion priced,” meaning the tolls fluctuated in order to limit use just enough to guarantee the free flow of traffic.
The original lease on the road included a noncompete clause that limited the state’s ability to add additional lanes to the non-P3 part of SR-91 or to build parallel infrastructure. This resulted in heavy congestion on the old lanes, pushing motorists onto the toll lanes and producing a financial windfall for the toll company. That ultimately prompted Orange County to buy out the toll company for $207 million in 2003.
However, the SR-91 problem is not inherent to P3s. It arose as a result of the conditions under which the franchise was arranged. Traditionally, P3s have been awarded through negotiations between private companies and transportation authorities, leading to high initial private investments and uncertainty about demand for the road. That risk, in turn, encourages toll road companies to want protections like the noncompete clause.
But there are ways to reduce the risk to the private companies while also protecting taxpayers and infrastructure users, as transportation experts Eduardo Engel, Ronald Fischer, and Alexander Galetovic explained in Regulation back in 2002. The authors suggest using a particular type of auction to award highway projects: a Present-Value-Revenue (PVR) auction. In a PVR auction, it is understood that the private company will only operate the road for a time, and then it will be returned to the public. Regulators set a maximum toll level that motorists will be charged to use the road. Private companies then bid on the amount of toll revenue, measured in present value terms, they would want to receive before the road is returned to the public. The lowest PV bid wins. The winner collects revenues and pays for the maintenance of the road until it has earned the revenue that it bid in present value, regardless of whether it takes five years or 50. Thus, the term of the franchise is variable depending on road usage; if usage is less than expected, the lease extends. If usage is greater than expected, the lease is shorter-term. The private company won’t “lose,” but it won’t make a windfall either.
PVR auctions provide resilience against shocks, such as lower-than-expected traffic. They also provide the basis for governments to buy back roads. For PVR franchises, a fair buyout is simply the difference between what the company has earned to date and the total revenue it bid to earn in present value.
In the case of SR-91, a PVR auction would have reduced the risk of the investment, and therefore would have precluded the need for risk-reducing contract obligations like the noncompete clause. If difficulties still arose, the government would have been better equipped to buy back the road for a more reasonable price.
Written with research assistance from David Kemp.
As part of its 2018 budget proposal, the Trump administration has introduced a plan to improve the nation’s infrastructure. The administration intends to reduce regulatory barriers on infrastructure projects and encourage greater private investment. It has also proposed increasing federal spending on infrastructure by $200 billion over 10 years.
A new Cato study provides input to the debate by examining infrastructure ownership and funding. Some people assume that the federal government plays the main role in infrastructure, but the states and private sector own 97 percent of U.S. nondefense infrastructure, and they fund 94 percent of it.
However, the federal government is the tail that wags the dog—its regulations, taxes, and subsidies affect the level and efficiency of state, local, and private infrastructure investment. The study argues that reforms to these federal interventions and privatization are the paths to higher-performance infrastructure.
Greater reliance on user fees, federal loans rather than grants, and corporatization are three keys to the Trump administration's infrastructure initiative released as a part of its 2018 budget. The plan will "seek long-term reforms on how infrastructure projects are regulated, funded, delivered, and maintained," says the six-page document. More federal funding "is not the solution," the document says; instead, it is to "fix underlying incentives, procedures, and policies."
In building the Interstate Highway System, the fact sheet observes, "the Federal Government played a key role" in collecting and distributing monies to "fund a project with a Federal purpose." Since then, however, those user fees, mainly gas tax receipts, have been "inefficiently invested" in "non-federal infrastructure."
As a result, the federal government today "acts as a complicated, costly middleman between the collection of revenue and the expenditure of those funds by States and localities." To fix this, the administration will "explore" whether transferring "responsibilities to the States is appropriate."
The document contains a number of specific proposals:
- Allow states to toll interstate and other federally funded highways;
- Encourage states to fix congestion using "congestion pricing, enhanced transit services, increased telecommuting and flex scheduling, and deployment of advanced technology";
- Corporatize air traffic control, as many other developed countries have done;
- Streamline the environmental review process by having a one-stop federal permit process and "curtailing needless litigation";
- Expand the TIFIA loan program and lift the existing cap on private activity bonds, both of which will make more money available for infrastructure without increasing federal deficits.
The paper also includes proposals for reforming inland waterways, the Power Marketing Administration, and water infrastructure finance. Like the transportation proposals, these call for increased reliance on user fees, corporatization, privatization, or loans rather than grants.
"Corporatization" means creating a non-profit or for-profit corporation that may be government owned but doesn't necessarily rely on taxpayer subsidies. Comsat is a classic example, but Canada and other countries' air traffic control systems work in this way.
Except for air traffic control reform, Trump's plan isn't fleshed out in detail. But these ideas have all been tossed around enough that everyone pretty much knows what they mean. Most importantly, they mean a significant change in the way Washington deals with infrastructure.
Because it doesn't contain a list of projects that members of Congress could take credit for, the plan has received relatively little notice in the media. Democrats, of course, are unhappy with it, but they would be unhappy no matter what Trump proposed.
One of the more controversial proposals is to allow the states to toll interstate highways. "I don’t like paying for a road twice," Representative Sam Graves (R-MO), who chairs the Highways and Transit Subcommittee of the House Transportation and Infrastructure Committee, told The Hill. But, given that Congress has had to inject tens of billions of dollars of general funds into the highway trust fund in recent years, what makes Graves thinks existing user fees are paying for the roads now? All roads need maintenance and occasional rehabilitation, so the fact that user fees paid for construction 50 years ago doesn't mean that costs stop.
The most important point is that Trump wants user fees to pay a greater share of infrastructure costs. Naturally, the transit lobby, which represents the most heavily subsidized form of transportation, per unit of output, is upset about this. But Trump's agenda sounds good to anyone who wants an efficient, user-fee-driven infrastructure program.
Last night’s address to Congress by President Trump was devoid of detail on infrastructure investment. But in justifying his desire to harness $1 trillion of public and private funds for “new roads, bridges, tunnels, airports and railways”, the President used two lines of bad economic reasoning sadly all too prevalent in public debate on this issue.
First was to invoke the building of the interstate highway system. “The time has come,” Trump declared, “for a new program of national rebuilding.” The implication: the interstate highway system was good for the economy, so we should invest more in roads today - a common rhetorical technique, but one which confuses average with marginal.
Previous economic research has indeed found that the construction of the interstate highway system substantially boosted productivity for industries associated with road use. But the same research finds those benefits to be largely one-offs, meaning this analysis does nothing to inform us about new decisions. In fact, more recent work has found that too many new highways have been built between 1983 and 2003, and that marginal extensions to the highway system tend not to increase social welfare, because the cost savings of reducing travel times are small relative to incomes and prices.
In other words, building a highway system can boost growth. Building a second highway system? Not so much. Rather than appealing to grand projects based on historical experience, all new government projects should stand up on their own merits – ideally having high benefit to cost ratios and being things that would not be undertaken by the private sector.
The second mistake was to highlight “creating millions of new jobs” as an aim or positive of any infrastructure spending. When the government is investing to build something, it should aim to do so most efficiently. “Jobs” in this sense are a cost, not a benefit, and ones “created” only come through the diversion of resources and opportunities in other parts of the economy.
Upon visiting an Asian country in the 1960s, Milton Friedman is frequently quoted as reacting to the absence of heavy machinery in a canal build by asking why the project was being undertaken by men with shovels. Upon being told it was a “jobs program,” he is said to have remarked: “Oh, I see. I thought you were trying to build a canal. If you really want to create jobs, then by all means give these men spoons, not shovels.”
If one is concerned with improving the economic growth potential of the economy, then you would base both the selection of projects and the means of undertaking them according to that objective. Sadly, when governments are involved, other ambitions (be it stimulating particular regions, appeasing certain interests, obtaining political prestige or facilitating observable jobs) tend to interfere with the stated aim. The constant talk of the benefits of wise, productive investment is an ambition, rather than something we should expect.