Thursday, March 01, 2007

Manville, on Why We Don't Use Congestion Pricing

Guest post by Michael Manville, UCLA.

Along with David King and Donald Shoup, I recently completed an article on the politics of congestion pricing, and Dave, Don and I are beginning another project on the same topic. Congestion pricing is getting a lot of press of late, and moving closer to reality, but politically it still has a long way to go. Rather than rehash our article here, I’ll discuss some of the broader issues about pricing’s political viability that I’ve been thinking about, some of which made it into our paper and some of which did not. I make no claims to completeness and only minor claims to originality. I also don’t want to implicate my coauthors in any of this, as they may disagree with some of it (although most of the good points are probably theirs). As a means of organizing the discussion, I’ll pose the simple planning research question that I think dominates this topic: if congestion pricing is so great, why don’t we do more of it?

I’ll put forward four possible explanations, none mutually exclusive but each with different research implications, and then end with a more general research question about the effect pricing’s political costs might have on the larger urban transportation system.

Explanation 1: No One Cares
The simplest explanation is that we don’t have much congestion pricing because congestion just isn’t a big problem. Yes, people complain about it, and the world might be more pleasant without it. But people will always complain about something, and if we go purely by preference revelation it seems that most individuals value access to unpriced roads more than they value time lost to traffic.

And speaking of lost time: since when does that require an intervention? People might waste a lot of time as a result of congestion, but are we really sure they wouldn’t be wasting time if they weren’t sitting in their cars? We like to think people liberated from congestion will be more productive at the office or spend more time with their kids. But they could just as easily use their newfound freedom to watch reality TV, eat trans fats, or blow the kids’ college savings on whiskey and blackjack. In surveys people regularly say that congestion is one of their biggest daily concerns, but that might just be evidence that most Americans have pretty good lives. Modern congestion relief is therefore just so much rain dancing and hand-waving. It doesn’t work but that’s alright, because no one really expects it to.

How convincing is this explanation? Not very. Preference revelation is a useful guide, but it can be deceiving. And the examples of the I-15 and SR-91 toll lanes (as well as burgeoning interest in other pricing projects) show that there is in fact some demand for speedier travel. Moreover, time costs are not the only costs of congestion. High levels of congestion increase the risk of accidents (although probably decrease the severity of those accidents) and therefore drive up insurance costs. Vehicles idling in traffic burn more fuel and emit more pollutants than vehicles flowing freely. And pollution costs, unlike time costs, do not fall primarily on drivers. Rather they fall disproportionately on those individuals, generally poor and many of them non-drivers, who live near freeways and other congested facilities. Pollution costs also fall heavily on future generations, in the form of carbon emissions and global warming.

Here we begin to see a problem, though. We have identified three groups that suffer from congestion: drivers, future generations, and the poor people who live near congested roads. One of these groups does not yet exist, another has very little political power, and the third—drivers—reaps most of driving’s benefits. This raises two issues. First, policies don’t spontaneously come into existence. Someone needs to get in the trenches and fight for them. But none of these groups seems likely to successfully organize and crusade for congestion pricing (I return to this point below). Second, clearly we have some distributional issues here, both across income levels and across time. And the current system seems to benefit the present and the affluent more than the future and the poor. Yet equity often considered a obstacle to congestion pricing, rather than a goal it could advance.

Explanation 2: Equity Considerations Prevent Congestion Pricing’s Implementation

Congestion pricing is regressive, so equity is a legitimate concern. The majority of tolls in a given metropolitan area are likely to be paid by the affluent, but some portion of the tolls will fall on the working poor.

But let’s parse the equity considerations some more. Tolls won’t fall hard on the poorest, because the poorest tend not to drive. The transit dependent poor may actually benefit from congestion tolls, because bus service will be faster and more reliable on tolled facilities. And even the poor who drive may gain from pricing, at least in some instances. The average poor person has a much lower value of time than the average affluent person, which suggests that a poor person will suffer a welfare loss by paying a market-clearing toll. But the average value of time in an income class is not the only metric of importance; we should also be concerned about the variability around the average. The fact that the poor do not on average have high values of time does not mean their time value does not sometimes see dramatic spikes. Rich people don’t have a monopoly on being in a hurry.

The equity picture gets further complicated when we consider pollution. Congestion pricing will reduce air pollution, and the benefits of that reduction will be greatest in the areas closest to the freeways. The residents of these areas tend to be poor, so the pollution reduction benefits of pricing will likely be progressive, even if the toll collection is not. However, pricing is likely to increase vehicle speeds, which will lead to increased noise pollution. Noise has some adverse health effects, but its major impact seems to be on home values. If most of the poor residents living near freeways are renters, the equity effects would be ambiguous: reduced air pollution, increased noise pollution, possibly lower rents (or rents that do not rise as fast) but also possibly reduced investment and maintenance by landlords. For poor homeowners living near freeways, a significant increase in vehicle noise could result in an erosion of already scarce wealth. And we need to keep in mind that equity benefits might not necessarily translate into political support for pricing. If the freeway-adjacent poor are drivers, and if they heavily discount the future, they may value the potential future benefit of decreased health problems less than the present and certain benefit of access to unpriced roads.

But it’s also worth wondering if we are being inconsistent when we single out tolls for being regressive. Most of the American transportation system is, after all, financed regressively. Local roads come from the property tax; highways from the gas tax. Transit systems are funded by sales tax increments, and transportation investments of all sorts are increasingly financed through bonds. Poor people pay regressive sales tax increments for commuter rail service they never use, and the outrage is…where, exactly?

So the fact that pricing is regressive may be a problem, but is it really a political problem? For better or worse, equity is rarely a decisive criterion in public finance, and equity is neither necessary nor sufficient for a program to achieve political success. Plenty of equitable proposals go nowhere, and plenty of inequitable proposals sail unhindered to approval. Is the equity-based opposition to congestion pricing just a smokescreen for people’s baser desire to avoid paying tolls—a classic case of “voting charitably and acting selfishly?

Explanation 3: Congestion Pricing Suffers from a Mismatch Between Economic and Political Efficiency
Congestion is an economic problem, and congestion pricing is an economic solution to that problem. But pricing also creates a political problem. Most drivers are voters, and vice-versa. Slapping a toll (and possibly a steep one) on the roads could alienate the majority in any political jurisdiction. Beyond that point, I can think of four inherent political weaknesses of congestion pricing

First, congestion pricing is complicated—a fact that often gets overlooked. Among economists and other transport professionals, the unanimity of opinion about pricing leads to the conclusion that it is obvious. And perhaps for economists pricing is obvious. But most people don’t think like economists and don’t share economists’ opinions about what is self-evident. For the typical voter, the only thing “obvious” about congestion pricing is that someone going to charge them to drive. Congestion pricing therefore requires explanation, and in politics, as the saying goes, “when you are explaining, you are losing.”

Second, policymakers tend to borrow ideas from each other. For this reason, congestion pricing suffers because it is still relatively rare. The prevalence of an idea both increases awareness of it and reduces the risk associated with it. This is true whether the idea is good or bad. For elected officials, there is political and psychological safety in adopting a widespread strategy, even a flawed one. It is better to make the same mistake as everyone else than to make a new mistake all your own.

Third, pricing places a restriction on an existing transportation benefit, while building a road or a transit system bestows a new benefit. Many attempts to alleviate congestion are notable for their emphasis on adding elements to the transportation system. Just as would-be dieters are told to “eat more greens” instead of “stop eating Twinkies,” political leaders tell us that we need “more transit”, or “more roads”, or “more bike lanes”—but rarely that we just need to drive less.

Fourth, congestion pricing is both local and transparent. This is its great economic advantage and its tremendous political disadvantage. Pricing reduces congestion because the people who drive pay the tolls, and they know the toll is charged because they are driving. Contrast that with road or rail construction, where the costs are buried and spread over all taxpayers. Most taxpayers don’t even know they are paying for these projects (how many Kansans knew they were helping build the Big Dig?) and even if they do know, the cost of the project is completely divorced from the behavior it seeks to influence. When everyone in the US is contributing a few cents to a light rail line in San Diego, no one has much reason to complain. But no one in San Diego has much reason to drive less, either. The opacity of road and rail subsidies makes them politically appealing, but also makes them less effective. The transparency of congestion pricing, by contrast, makes it effective, but also creates a huge political barrier to its implementation. Which leads to a final explanation:

Explanation 4: Congestion Pricing Lacks a Powerful Advocate
This is the point we make in our paper. If we leave aside the toll revenue, in most American metro areas the distribution of congestion pricing’s costs and benefits will make more people losers than winners. Further, many people who do not lose from congestion pricing might think they lose. The problems this creates are twofold. First, people might be actively opposed to pricing. And second, people who aren’t actively opposed are unlikely to be actively supportive; they will be indifferent. This indifference creates a collective action problem. Indifference is helpful once a policy is in place (no one tries to overturn it), but a serious problem before it is implemented (no one fights to get it started). People fight to implement programs that offer them large benefits. But to whom does congestion pricing offer a large benefit? Who will step forward and secure the initial approval of congestion pricing?

This is where the toll revenue comes in. Congestion pricing has the potential to create piles of money. People like money. So it seems fair to say that whoever gets the money will also like congestion pricing. But some candidates for the toll revenue are better than others. A good revenue claimant has a combination of strength and benevolence: strong enough to lobby successfully for pricing’s implementation, but viewed by the public as something more than just a profit-hungry rent-seeker. In the right hands the toll revenue can be a “policy ratchet”—a preference-altering instrument that helps secure pricing’s approval and reinforces support for it once it is in place. But not all claimants will generate ratchet effects. Drivers will not. They are dispersed and disorganized, and I think buying them off with toll revenue is a losing gambit. The key, again, is to use the revenue as lever to get pricing approved. We (my coauthors and I) don’t see drivers pulling that off. We also don’t see transit agencies doing it.

In our paper we argue that local governments are the best claimants for freeway toll revenue: they are effective lobbyists at the state and national level; they are generally seen as benign (with big city governments being an exception); and, because they provide ground-level basic services, they tend to be attuned to their residents’ preferences. So we argue for a system where local governments spend toll revenue to augment public service provision. This gives us powerful claimants who could nevertheless put the money to visible public use.

Giving the revenue to local governments also lets revenue distribution work with, rather than against, fragmentation and Tiebout-sorting: wealthier communities could spend their portion of the revenue differently than poorer communities, and we wouldn’t end up trying to find one or two region-wide programs that everyone can agree on. I think pricing will have a better shot at acceptance if a heterogeneous region can spend its toll revenue in a variety of ways, rather than being told it is all going to transit or new roads. If the tolls are perceived as benefit taxes rather than transfers, their political acceptability might rise. Others have argued differently, however, and this question is far from settled.

Congestion Pricing and Efficiency
One obvious research question is whether any given congestion pricing program meets efficiency criteria. (See here and here for different takes on London). But my question is somewhat different: Would the widespread adoption of congestion pricing make the overall urban transportation system more efficient? My inclination is to say yes, but I can think of reasons to say no. Congestion pricing will unhide many of the costs of driving, which is no small thing, and it can make existing transit systems more efficient.

But pricing has high upfront costs—both political and capital—and the presence of such costs always comes with the incentives to hide them. Capital costs get shifted to non-users, and political costs get buried in logrolls and other deals. The federal government, for instance, paid the start-up capital costs of Southern California’s I-15 toll lane. Singapore's program was preceded and accompanied by quantity regulations, the I-15 sends much of its toll revenue to an underused bus line, London purchased 300 new buses when it introduced its cordon charge, and so on. Those of us who favor congestion pricing tend to think of it as a substitute for other, less efficient policies, but we are unlikely to implement congestion pricing by itself; rather it will arrive, for political reasons, bundled with other policies whose merit and efficiency varies greatly. So the possibility exists that pricing will deepen, or at least not drain, our current swamp of subsidies.

I would end right now but someone is probably saying, “Why hasn’t he mentioned privatization?” And the answer is that I don’t know much about it. But, yes, if private companies build roads just to toll them, some political distortions can be avoided—the drivers pay, pure and simple. In the abstract, that makes some sense. My stab at a research question is this: just how private are the private roads likely to be? If a private company wins a grant or tax credits to defray its capital costs, some of the efficiency gains go out the window. Ditto if the private company exacts some sort of noncompetition clause from the public sector, as happened in Orange County. And even if none of this happens, should we be building roads just to toll them, i.e., increasing supply purely to better manage demand?

On efficiency grounds, doing this would require an a priori assumption that more roads are necessary, and that the proper pricing of existing roads would not create sufficient capacity on its own. If we reject this assumption, does the new road count as a politically-introduced distortion? After all, it is being built because the first-best option—pricing existing capacity—lacks the necessary support. Then there is the question of where these new roads would be located, especially in dense areas. Will these roads require public intercession in the form of eminent domain, and if so, what are the various political trades that might be required to enable that?

None of this is to say that private road building is unnecessary or undesirable, only to wonder whether the necessity and desirability is more political than economic.

The Bottom Line
For most of its intellectual history, congestion pricing has been the domain of economists and engineers. But now it seems, at long last, to be on the cusp of widespread implementation; the last obstacle is the political one. This gives planners, who know something about the messy politics of policy development, an opportunity to influence the way it is implemented, and to offer explanations if and when its implementation goes awry.


[p.s. Editor's helpful video addendum: An example of a congestion fee implementation and a translation of economic thinking for the uninitiated.]

No comments:


My blog has moved!

You should be automatically redirected in 6 seconds. If not, visit
and update your bookmarks.