Michael Munger on Traffic
Apr 2 2018

congestion%20pricing.jpg Does rush-hour traffic drive you crazy? Is a congestion tax on car travel a good idea? Michael Munger of Duke University talks with EconTalk host Russ Roberts about the economics of traffic and congestion taxes. It takes a while to get there (how appropriate!) but they eventually agree that a tax on congestion while reducing travel time is harmful to many drivers and may be best thought of as any tax placed on a particular good--a way to raise government revenue from the pockets of the consumers of that good.

Bruce Yandle on the Tragedy of the Commons and the Implications for Environmental Regulation
Bruce Yandle of Clemson University and George Mason University's Mercatus Center looks at the tragedy of the commons and the various ways that people have avoided the overuse of resources that are held in common. Examples discussed include fisheries, roads,...
Michael Munger on Permissionless Innovation
Michael Munger of Duke University talks with EconTalk host Russ Roberts about permissionless innovation. Munger argues that the ability to innovate without permission is the most important concept of political economy. Munger defends this claim and explores the metaphor of...
Explore audio transcript, further reading that will help you delve deeper into this week’s episode, and vigorous conversations in the form of our comments section below.


Allen Hutson
Apr 2 2018 at 9:54am

I think that you guys may have missed another point in your discussion at minutes 15-20.

If drivers are taxed at 5:00, the people with the lowest value of their time have an incentive (along with everyone else) to enter the highway earlier. But because their time is less valuable, they will be more flexible about when to enter the highway. those with higher valued time will be less flexible about when to enter the highway.

The system literally creates an incentive for users with lower time values to game the system by entering the highway at 4:57. So on average, you will in have users with higher time values paying the toll and those will lower values of their time free-riding.

Lee O'Neill
Apr 2 2018 at 9:56am

Russ, I raising the question of morality in connection with congestion pricing is exactly the right place to go but the universe in which you raise it is, I think, too small. It isn’t a matter of suggesting an immoral transfer from automobile drivers to some arbitrary third party but a question of how tax money overall is allocated. Every dollar (from whatever source) spent on roads or parking lots or facilitating gasoline production or whatever, is a dollar not spent on mass transportation (or any other worthwhile goal of the society as a whole). Ideally, an analysis of the traffic situation would include a wider analysis of other modes of transportation. After all, don’t the unpleasant conditions of most subway systems, on-street parking that makes bicycling dangerous, and confusing bus systems contribute to automobile congestion? Should the opportunity costs of parking lots, garages, and on-street parking that facilitates the storage of cars when not in use and air pollution be considered as contributing to the cost of driving?

Michael C. Munger
Apr 2 2018 at 12:59pm

Allen Hutson: That’s precisely, verbatim, the point I was trying to make in this section:

Michael Munger: There’s a bunch of people already in the central city. That’s what’s causing the congestion. We’re going to trigger congestion pricing once there’s already congestion. That means that some of the people already there who didn’t have to pay, have a low opportunity cost of time. So, you’re saying–

Russ Roberts: Some might have a high opportunity cost of time. It’s mixed.

Michael Munger: Yes. But even the ones that have a low opportunity cost of time, it’s already congested. Because we didn’t charge them.

Now, I wasn’t very clear. But you are right, and that’s what I was trying to say. It’s an interesting and (I think) important point, so thanks for making it clearer than I did….

Dan Holstad
Apr 2 2018 at 1:05pm

I don’t understand why you think privatization would be better than a congestion fee. It does not solve the compensation problem, both parties are still worse off. Now instead of having some other government service funded, they are giving their money to a rent seeker. While there is a chance that the person who can no longer afford to use the road may get some use from the school there is almost no chance that they will benefit from making the private corporation keeping the money from congestion tax.

Bob Pierce
Apr 2 2018 at 4:05pm

I am not an economist and I have a very thin grasp of the things talked about in this episode.

I will say that early in the episode I thought the point was that the congestion tax is not an exchange. The driver who values the use of the road more greatly does not pay the driver who stays at home (or gets off the road) after 5:00 pm. I thought he point was something like, it would be best to not only penalize the drivers heading into Manhattan, but to also allow the ones who leave to be compensated. Sort of a dual model. Create incentives to, at the same time, discourage drivers from heading in and pay other drivers to head out.

So, I thought for about the last 45 minutes of the show that you both were driving to a sort of “aha” moment where you were going to say “Yes, but there could be an internet application that matches up drivers and allows the drivers who value entry into Manhattan at 5:01 pm to pay a driver to get out of Manhattan.” Or something of that sort. An auction or something.

But I was disappointed at 1:03:39 to hear Michael say that something to the effect that matching up drivers like this is impossible.

Could there be a lack of vision here? What if such an auction application were possible? Better?

Philip L.
Apr 2 2018 at 4:48pm

I think there are some far-out technologies that may allow the sort of truly private road structure that Mike talks about at the end of the podcast. The first may be The Boring Co.’s tunnel systems that they’ve began developing. So far, these are totally private enterprises, funded by branded flamethrowers rather than tax dollars. The second may be flying cars, which may be only a couple decades out (once we solve the nasty problem of untrained human pilots. . .)

Moshe Schorr
Apr 2 2018 at 5:02pm

Great episode. One point I found strangely missing:
Congestion pricing isn’t just about allocating this resource among drivers — it’s not always — in fact, usually not, a road in the middle of nowhere. It’s usually an urban area. Despite never having driven or ridden in a car in downtown London, I benefited from the congestion charge as a pedestrian. I benefited from the cleaner air, from the more pleasant streets, and so on and so forth. So there’s more than just the drivers to think about, and also, I’d think this means, arbitrarily making it scarcer is potentially beneficial.

Jacob Karasch
Apr 2 2018 at 5:22pm

If the “good” being transferred around is “restraining from being on the road” it will be problematic to have a market solution because restraint is impossible to measure. A person who isn’t on the road must WANT to be on the road if they can qualify as restraining themselves.

It is much easier to look at the road as an asset owned by the taxpayers. If, facilitated by technology, you paid according to the time your car is on the road, and the money was given equally to all taxpayers, you would create an effective market. This technology, however, is problematic for several reasons. Perhaps though, you could use a gas tax as a proxy. The proceeds from gas tax go to the people. This way, it doesn’t disproportionately harm the poor as they are beneficiaries of the proceeds. This increases the cost of being on the road and pushes people to ride bike, use public transit, work from home, live closer to work and travel off-peak. Those that NEED to travel (for some lucrative opportunity or to get to the hospital) are able to do so.

Now everybody write you local congressman that you want a $10/gallon gas tax. lol

Chase Steffensen
Apr 2 2018 at 5:43pm

How did Coase not come up in this discussion?

Jeffrey Klein
Apr 2 2018 at 9:04pm

To second Lee, I think this episode missed a bit by being overly narrow and deep. The question of whether or not congestion pricing is a net benefit to drivers is a somewhat interesting economics question, but those advocating for congestion pricing aren’t particularly concerned about their convenience (after all, American society has spent the last century catering to their convenience almost exclusively). A congestion tax is the price on the externalities placed on everyone else in busy city – the pollution, space, danger, misery, and damage caused by private vehicles in places crowded with pedestrians, cyclists, or people just trying to enjoy a cup of coffee on a sidewalk table.

I would be curious to know if you support pricing parking on public streets, where private automobiles take up a disproportionate amount of space, as they do when in motion. While it may fail to be a true market, it still strikes me as wise to charge rental for the space, and it’s not even that hard to get the price right.

Adam Cozzette
Apr 2 2018 at 10:47pm

Russ and Michael: if I understand correctly, you both claim that congestion pricing is immoral on some level because it involves the government stepping in and taking something unfairly from one group of people, namely low-income road users or those who otherwise have a low value of their time. You say in particular that this is a net loss to drivers as a whole.

I think you’re missing an important point, though, which is that most roads do not pay for themselves and are propped up with general-fund subsidies. Last year the Tax Foundation came up with estimates on what percentage of road spending in each state is actually covered by road users: https://taxfoundation.org/state-road-funding-2017 For example in my state, California, they found only 61.8% of road funding came from tolls, user fees, and user taxes, so about 38% came out of general fund sources. In other words, to a large degree we’re all forced to pay for roads no matter how much or how little we use them. So I think your claim is indefensible–you can’t object to congestion pricing taking something from drivers if you also just ignore the fact that the whole road funding setup is based on taking money from the general public and using it to subsidize driving.

Toward the end of the podcast you talked about privatization of roads. If I remember right, one of you pointed out that this almost never happens because no private company has figured out how to make money doing this. This hits the nail right on the head–private companies have to turn a profit or else they go out of business. Only the government has the ability to lose tons of money on roads but make up for it through other funding sources.

Apr 3 2018 at 12:16am

Thank you for a great podcast. Some thoughts for your consideration.

Following on Dr. Roberts’ point about morality. It is important to note the difference between costs and waste. Waste is excess costs or spending minus cost to cure. While cost alone does not factor in the counterbalancing expenditure to remedy.

When Piguovian Taxes are applied social planners attempt to be agents for the aggrieved and charge the cost of the externality. Cost, however, is not waste. When charging for externalities two adjusting factors come into play:
1. The market reduction in supply due to the increased cost, and
2. The entrepreneurial solution to reduce the costly factor.

My assessment is that the reduction in supply is very limited (between 5% and 10%). The entrepreneurial effect can be large but is not predictable. So, if no material innovative means of reducing the externality exist, substantial costs to the aggrieved remain.

One should note that the collection of the externality tax will not improve the welfare of society by the amount of externality tax, but only a few percentage points of the externality, before accounting for transactions cost.

Transactions cost can be substantial and in the congestion externality game even with smart tags to collect tolls the collection cost can exceed the value of the reduction in congestion.

I am concerned with this because the assignment of the tax tends to assign full value of the externality tax to the social planner (government). I submit that only a portion between zero and the amount of the externality reduced minus the transaction cost is fairly earned by the social planner.

It is reasonable to assume the transaction cost is both paid by a portion of the externality tax and paid out in full without undue profits.

The other portion of the externality tax must be re-introduced into the economic system without the power of the market knowledge. I think this is what Dr. Roberts was pushing at in the beginning of the podcast. General equilibrium factors lead me to believe that gifted resources have less direction because they were not earned for a purpose. The gifted return will tend to reduce the marginal willingness to work. Gifted resources may be re-introduced at a socialist efficiency rather than a market efficiency because of the lack of market format.

I am aware in perfect competition that the theory is that this excess revenue can be efficiently return to the economy, but in reality, tax tend to have concentrated benefits and disbursed costs (externality taxes) that would can and likely will add more distortions to the economy re-inserted the externality revenue than the corrections made by the tax.

I note the disingenuous of the social planner agent. If the planner is representing that the aggrieved is harmed, then the residual claimant of the funds the social planner collects on the aggrieved behalf should be the aggrieved.

As discussed in tangents before, but not directly the aggrieved takes remedial actions to lessen the externality when the aggrieved property rights are limited. Assignment of the full property right can provide significant distortions as Coase effectively points out.

I would suggest that the current property right system is that the aggrieved has the property right to collect tort damages from externality producers and this limits the damage. But the collection the aggrieved needs to incur to collect reduces the supposed producer of the externalities liabilities so the property rights are closer to 80% producer and 20% aggrieved. This is a feature not a bug and prevents aggrieved from filing suit on trivial levels of annoyance.

It was with great pride that Dr. Munger mentioned my proposal, but I present my version here:

Herbert Mohring figured this in 1962. The toll/user fees should be capped at the amount that the marginal benefit of more system equals the marginal cost of the benefit. This introduces the potential to contract for improved roads and streets with the proceeds of the tolls. I believe with the mindset of improving users experience many potential projects are possible. I am also for virtual pareto optimality 90% of the users need to see material benefits or the program is configured incorrectly. Payment is best made to those that would have been forced off the road by providing 10% more road with the revenues so that they can stay on the road and travel a little faster. This method properly compensates those that would have been harmed by an increased toll.

I am less convinced that this is the approach in practice now.

Dr. Munger correctly ascribes my belief that NY is one of the best places to provide a congestion tax for all vehicle, but it is still questionable if the benefits outweigh the costs.

Thank you both for your work.

Apr 3 2018 at 8:59am

If I’m understanding right, I think Russ’s claim that a congestion tax can’t make everyone better off depends on this being an analysis of a one-time choice. If we think about using roads day after day, it is possible that everyone could be better off if their net benefit from reduced congestion on some days is more than their net loss from the tax on others.

And to push this a little further, there’s an insurance benefit: maybe one day I’ll need to ride in an ambulance through an otherwise congested area, maybe not, but ex ante there’s some benefit to the fact that there will be less congestion if that happens.

Don Crawford
Apr 3 2018 at 11:23am

Roads are not free. We pay for them in gas taxes and general taxes now. But there is no connection between supply and demand such that the more that is paid the more the supplier is incentivized to produce. Congestion on our roadways is just like the lines outside grocery stores in communist Russia–it is simply a result of the government interference in the market.
Instead of taxes we should pay tolls to a private roadway company. What has to be sold to the private company is the route and the right to charge for people using that route, so the private owner has the incentive to build more roadway in that location and charge tolls for using it. A private company owning the roads and taking tolls (as you move from one EZ-pass type reading device to the next) means the company is properly incentivized to increase the supply of roadway. They are also properly incentivized to control entry (through dynamic pricing) such that traffic keeps flowing–and in fact the speed of traffic should control the price. They won’t make money if traffic is at a standstill.

Yes, it’s expensive to create additional roadway in places like Portland Oregon where the main north-south freeway was built 50 odd years ago in a very narrow space with no room for extra lanes. However, that is the opportunity to make money from the provision of extra roadway where people want it. The market will come up with ways to double and triple deck roads. The more people that want to travel in that particular area the more money there is to be made and the more roadways that will be built.

Why assume the roads will just fill up? That’s like saying there’s no way there can ever be enough chickens for everyone. Whenever there are more chickens in the butcher shop they just sell out!
There are a finite number of people so it is possible to build enough road capacity to handle them, again assuming pricing is correct. Now maybe it will turn out that the actual cost of providing roadways for individual cars if it were born by the individual drivers directly would make alternative means of transportation more desirable. But “mass transit” needs to be priced at its actual cost at the same time, not be subsidized by taxpayers who aren’t using it.

The cost of traveling the road paid by users MUST go to the entity that is providing the supply of roadway. That is the only way that the law of supply and demand can work correctly. Yes, a “congestion tax” that goes anywhere but into road-building is not really a good idea.

Don Crawford
Apr 3 2018 at 11:42am

I said in the above comment that “the speed of traffic should control the price” which maybe isn’t obvious. When the roadway is congested the traffic slows down and that’s the time for higher pricing. Probably would develop dynamic signs displayed at entrances so people could choose not to enter the roadway when prices were too high. Seems to me that the data showing how long it takes to get from one EZ-pass type reader to the other would inform the system what the price ought to be. Heavily traveled corridors would have higher tolls and the opportunity to double the number of cars paying tolls would lead the private owner to add a level of roadway.

Apr 3 2018 at 12:56pm

Both Russ and Michael do an excellent job of distinguishing congestion taxes from a market. The tax doesn’t compensate those who choose to avoid times of congestion.

Yet they both seem to end up with the dismal conclusion that the situation might not be amenable to something better. For example:

Russ: “So there’s no attractive … My conclusion to this is that I think you just have to live with the fact that when a lot of people want to live near each other there’s going to be congestion.”

Why not have an arrangement in which those who travel at low congestion times get compensated by those who want/need to travel at other times?

1. Instead of designating lanes for high occupancy, designate lanes for cars employing some type of “EasyPass” technology. (Over time, the number of such lanes could be increased if this become more popular.)

2. Without any regard to how many people are in the car, if a car travels the EasyPass lanes during above average congestion for those lanes, they pay. Possibly the more congested, the more they pay. (High congestion for the EasyPass lanes can still be less congested than the no-charge non-EasyPass lanes.)

3. Cars that travel the EasyPass lanes during average congestion for those lanes are not charged.

4. Those cars that travel the EasyPass lanes during below average congestion for those lanes get a credit to their EasyPass account. It would not need to be a large credit. It shouldn’t be so much that it becomes a significant source of income. (It still costs gas and time and wear on the car to travel.) It simply tips the balance such that those who can drive at uncongested times have an extra motivation to do so. The difference in cost compared to high volume times compensates their choice to use low volume times.

5. Through apps or websites or other means, make known present conditions of congestion and patterns of congestion over time. (Apps like Waze already take driving congestion into account for selecting routes. Such algorithms could take EasyPass variable pricing into account.)

Wouldn’t this create more of a market arrangement in which money can flow from those who want to travel at certain times to those who are willing to travel at other times?

Gene Epstein
Apr 3 2018 at 1:40pm

I believe it clarifies matters to think about the “congestion pricing” practiced by owners of bars, long before the idea got devised by economists. It’s called the “Happy Hour.” Drinks are cheaper if you order before 5:00 (the Happy Hour); then drinks go up in price after 5:00. This must have all started because the bar-owners noticed that the bar is jammed after 5:00 (congested), and that they can make more money by hiking prices after 5:00 and cutting prices before 5:00.

Do Russ and Mike Munger object to this practice? If so, on what grounds? People who can get to the bar before 5:00 benefit because they pay less for a drink; and people who go after 5:00 benefit because the place is less crowded when they patronize the bar, and are clearly willing to pay the price.

If roads and streets were privately owned, then business owners would no doubt use the same “Happy Hour” approach–maybe they’d dub the policy “Convenience Hours Pricing.” Russ objected to privatization of roads and streets because he apparently believes this would put undue monopoly power into the hands of business owners. But that’s a different matter. Regarding the main issue, since every user of the privately-owned road or street is charged a fee to begin with, wouldn’t the discounts and premiums make most users better off,just as Happy-Hour pricing makes most patrons of bars better off?

And by the way, it was Harvard’s Ed Glaeser who pointed out the virtues of charging only users–not taxpayers in general-for infrastructure such as roads, streets, and bridges. Government could do the same.

Apr 3 2018 at 2:20pm

Lot of the podcast is spent on arguing how congestion pricing is not a market. I agree that the arguments show that it’s not the same market as one would get if (barring transaction costs etc) those valuing their time highly would be buying less congestion from those on the road valuing their times less. However, this doesn’t mean that it isn’t some other kind of a market – in fact, congestion pricing amounts to a monopolist (the government) selling to would-be drivers. That those who choose not to travel because of a congestion tax do not get compensated doesn’t sound to me obviously different than the fact that one won’t get compensated for not going to highly-priced rock concerts. At any rate, the show did little to convince me that a benevolent monopolist should set the price of driving on a congested road to 0.

Steve Wood
Apr 3 2018 at 3:40pm

Russ Roberts opened the door to confusion on the part of listeners and well, I have to admit that I didn’t hang in here as much as I’d have liked …

I wonder if there could be congestion tax that requires no toll or money at all (except perhaps for the cost enforcement). Would a suitable congestion tax be to reduce the speed limit on highways during rush hour?

During the peak driving periods, lower the highway speed to 45 MPH or even lower, which would effectively be a tax.

Unexpected benefits might be that the traffic patterns would be more uniform, with fewer starts and stops. And conditions could be safer, leading to fewer accidents, fewer lane closures, rescue/wrecking equipment, fewer jams, greater flow/continuity. (Fewer interruptions may actually equal higher net speed anyway, so travel time may not be increased after all.)

Of course, this “tax” may not work inside an island (Manhattan) but it could in cities that feature interstates, loops, expressways.

David In Austin
Apr 3 2018 at 5:45pm

To return the toll money to the drivers that don’t drive during the peak hours, use the toll revenue to reduce the car-registration fees and gasoline taxes in the county where the taxes are generated. If putting the money into the roads isn’t practical, you could add services like staging tow trucks for expediting “incident clearance“.

Even under the clearly bogus model that everybody values their time the same, you can still get benefits overall because of the non-linear nature of congestion. In some conditions it’s possible to double the average speed by just taking 10% of the cars off the highway, for example. The throughput of cars on a highway can actually increase when you remove some cars.

Not only do people not value their time equally, individuals have different values of time across days. If you happen to be late to pick up kids at a daycare, or are riding in the back of an ambulance with a heart attack, you probably value time by a factor of 10 or more than usual.

Even ignoring all that, people in higher-capacity vehicles (car pooling or bus-riders) wind up paying less per person in congestion fees so they would see a surplus as well. Less congested roads might make up for the time lost in finding a car-pool mate, or taking a bus.

Apr 3 2018 at 7:18pm

Another interesting episode.

I have to say that there does seem to be a simple, if imperfect, way to narrow down on how much the trip is worth to someone that is not merely a rehashing of how much money they have (the pay-to-play concept). If it is a flat $5 fee to travel a mile between 5 and 6, the rich don’t care and travel anyway, the middle-class sometimes travel and sometimes don’t and people earning minimum wage don’t travel during that time. That reduces traffic, but with the perverse effect that you have hourly workers getting off work at 5 who hang around an extra hour because they don’t want to spend $5/mile to travel home.

On the other hand, if there is a system that differentiates how much money people have in the first place and charges them different amounts, then everyone thinks about it and the people for whom taking the trip at 5 is significantly better than taking it at 6:30 leave at 5. Since a system requires that each car is tracked, and the government has an estimate of what car is worth already for property taxes purposes, a relatively painless proxy of a person’s wealth is the value of their car. Making the tax a percentage of the value of the car rather than a flat fee makes people consider whether traveling at the desired time is actually worth it. If it costs 0.02%/mile, someone with a $50,000 car is paying $10/mile while someone with a $20,000 car is paying $4/mile and someone with a $4,000 car pays $0.80/mile. There are plenty of imperfections with this system, but it addresses the biggest sticking point I have with taxing roads at heavily congested times: if roads are public goods (which they are), excluding poor people who have no backup option makes their lives significantly worse off, while wealthy drivers throw some change in the tax pile. In cities where public transportation is adequate, this is a smaller problem, but those cities are few, so oftentimes the next best alternative to driving at peak times is far worse.

Presumably this could be worked out in a way that means that property taxes on cars can be reduced; in effect, people who do not use the roads at peak times would then be paid to not be doing so (relative to before any tax was instituted) while people who pay to use the roads at peak times would pay the same as before or significantly more than before, all based on their desire to drive at peak hours.

Joe Gordon
Apr 4 2018 at 1:19am

Many great comments here, and a fun podcast with a few things over my head, and a few things learned.

I think there is already what I would (inappropriately, I’m sure) call a “natural tax”. I’m a contractor, and I travel all over my region to get to jobs. I have to account for traffic. If I’m late, I might lose the job, or look worse to the customer (losing word-of-mouth future customers). I’d also be late to the next job that day, and the one after that. Or maybe, since I’m late, I hurry the job and do sub-par work. All of these things lead to negative consequences. So I have to schedule my jobs so I get there on time. I either schedule them such that I don’t drive in traffic, or if I know that I will have to drive in traffic, I leave early to compensate.

I live in Sacramento, the capital city of California. We have thousands and thousands of state workers downtown, and many of them are fortunate enough to have flexible work hours. I know some who get to work at 5am and leave when other people are going to lunch. Others do long days, and only work four days a week. This allows them to avoid traffic and not contribute to congestion. I don’t know of government-sponsored incentives for corporations to do this, but since the state bears the cost of the impact of congestion, they (the state) benefit by giving these considerations to state employees.

Luke J
Apr 4 2018 at 2:37am

Interesting conversation, but I think the widget example is the wrong starting place when tackling roads as no two drivers are exchanging a good produced by the other. Even if we stretch to say the “widget” is the place on the road, (i.e. Driver A pays Driver B to vacate), this has near zero effect on congestion.

The comments by Adam Cozzette and Don Crawford are spot on. We’re not really interested in efficiency, we’re interested in shifting more costs of providing & maintaining roads onto the people who use them. Better a user fee than a congestion tax.

Apr 4 2018 at 7:56am

I’m a bit disappointed no one raised the idea that in certain high-density cities (Manhattan part of NYC, Boston, London), at certain times of the day, driving a car is a luxury good, substitutable with public transit. We already differentiate between people with low value of time/poor people and high value of time/rich people on this basis. Obviously, if you live in Oklahoma City, driving is not matter of choice for anyone who works.

Throughout the podcast, I understand the point that a tax just takes away money without giving anything back, but Munger raised the point, which was never visited again, that this could be spent on improvements mass transit. Win-win for everyone. Rich people/high value of time people keep driving with less congestion. Everyone else gets a better commute time than they would through an improved mass transit system that has economies of scale working for it. The only argument against this is the middle class people who now don’t feel rich enough to drive anymore and so choose mass transit instead. I can see the issue a libertarian (US) would take in that these people are being coerced to new actions. But, everyone’s already paying a tax to build and maintain the road in the first place. It’s just a question of whether that tax adequately captures the true cost (including cost of congestion) of maintaining and using the road.

Don Crawford
Apr 4 2018 at 2:11pm

The money we taxpayers shell out for the roads, partly in the form of gas taxes, in theory goes to the provider of the roads–the government. But these days the government takes that money and spends it on things like light rail or bike lanes or other things. Even when spent on roadways government contracts are very expensive with all the rules about prevailing wage etc. Around here governmental officials believe that their mission is to decrease our access to roadways and to push us into mass transit or onto bicycles. That is the root of the problem.

We need private ownership of roadways and user fees to have a correct relationship between customers of the roads and providers of the roads. With private owners and tolls the provider has the incentive to increase the supply to meet the demand and enable more customers to use the roads and to make more money thereby. I assert that congestion would be a more rare event if this were the case. Sometimes they run out of my favorite tortillas in the store, but in general I can get what I want when the market provides it.

Craig H
Apr 4 2018 at 2:49pm

Private owners of roads are rent seekers and will not invest in an increase of the supply of roadway. To suggest otherwise is absurd.

Dale Eltoft
Apr 5 2018 at 12:14am

I propose that the idea of using money as a cost mechanism to allocate road usage is based on a questionable premise. Roads are a public resource and therefore we all deserve equal access. However, we do not all have equal access to money. Therefore if we want a true market the government should allocate equally to all of us a fixed amount of “RoadCoins” (maybe a cryptocurrency). Then set up a marketplace for selling them.

This doesn’t solve the problem of how to set congestion pricing but it at least it fairly allocates the public resource.

Apr 5 2018 at 10:03am

Dale, I like the idea of using a cyptocurrency to set a market for roads. However, I do see a few ways in which it might fall short.

First, unless excess “coins” can be converted into real money at some point then people who choose not to travel at peak times never get compensated for it, which was discussed some on the podcast and I think is an important point. For example, there are plenty of people for whom it doesn’t really matter when they go to and from work (typically white collar jobs that don’t have days full of in-person meetings), so they should go to work at 9:30 and leave at 6:30 (if rush hours are 8-9 and 5-6) or go at 7:30 and leave at 4:30; should is in the sense of “it is best for society for them to do this”. For someone like this to be incentivized to change their schedule a little (beyond avoiding the pain of traffic, but those people already adjusted their schedule) they need to be compensated. This can happen either with giving them actual money, or having them not spend actual money; not spending cryptocurrency that has no value aside from being used for roads does not adequately get them off the road.

Another point, which is fairly secondary in my mind to the first one, is that we arguably do want high-earners to travel whenever they want and not be limited by their limited ration of road coins. Theoretically they get paid so much because they bring so much value to society, so we do not want to hinder them from traveling at the best time possible for them and thus reduce their productivity. I would posit that a mere pay-to-play scenario would be unfortunate, which is why I suggested a toll based on the value of one’s car a few comments above (in short, nobody is priced out if they really want to travel and everybody has to at least consider the cost, aside from the ultra rich). To the point of roads being a public good paid for by everybody, without delving into the philosophy of it too much or taking much of a stand, I will note that the high earners did technically pay for more of the road initially than lower earners, although that certainly does not mean that lower earners should not be able to use it.

Mike Williams
Apr 5 2018 at 12:00pm

Re: examples of offers to build private roads.

At least one developer has offered to replace Toronto’s Gardiner Expressway (a very congested arterial into downtown closely paralleling Lake Ontario) with a tunnel. The one in the 90’s didn’t even want to make it a toll road, they just wanted the above ground city-owned land that the current road occupies (it would have been a privately funded version of Boston’s Big Dig). The City council turned it down because they didn’t want someone profiting from roads.

The private road in Toronto mentioned in the podcast is the 407. It was built as a private toll road in what was a the time unused land North to Northwest of the city (it initially wasn’t an ideal stretch of highway to use but its utility has improved over the years). It is a good example of congestion-pricing since it is expensive to use but rarely busy whereas the alternatives are free but typically very busy. And once they complete the extension allowing you to bypass Toronto entirely, you will be able to pay a hefty fee to avoid spending time in traffic on the public/free 401. It will be interesting to see when the 407 is a true substitute to the 401, what value saving 1/2 to 3 hours for a cross-Toronto trip has.

Kristian Lande
Apr 6 2018 at 8:48am

You already have a congestion market for traffic into and out of NYC and it works incredibly well. The problem is that you have to be an electron. Seriously though, congestion markets work in electricity and they would work with traffic and you have the technology to do so. You would pay people some amount to relieve congestion (driving out of the city) thus freeing up parking space and you could even create a market based on real-time pollution monitoring. Electric cars would pay the congestion charge but receive some market set rebate for relieving pollution.

Jakob Engblom
Apr 6 2018 at 12:57pm

Many different points where made in the podcast, some really interesting.

To me, the key take-away for the future is the obvious that prices alone does not make a market. and that indeed a congestion CHARGE is a not a price, but a tax or public charge. Principle applies in other places too.

The eternal admiration for Uber still strikes me as fundamentally incorrect. They are basically subsidizing prices using VC money to drive regular taxis out of business, and putting the capital costs onto the drivers instead of themselves. All rather immoral and “bad”… A simple deregulation of normal taxis sounds like a better solution to me – same rules for all that do the same thing. Let regular taxis have freer pricing.

A key point also made by other commenters is that congestion charges also benefit other people. In particular those that live in the city and get less traffic messing up the place, and deliveries and taxis that can move faster. As well as nudging people towards public transport. The cost of the charge in Stockholm is very low compared to typical earnings, so it is not like someone could not afford it if they really want to.

Tom N
Apr 6 2018 at 3:35pm

To make congestion into a market, as you hoped for, could be done (I think) with a new feature upgrade (and a bit of user education) inside apps like Waze. If you use those apps, then they know where you’re driving to and about how long it will take. If you tell it a bit in advance where you’re going and when you want to get there (or when you want to leave), it could add in a “pay / get paid” option. If you want traffic to decrease, you could offer an amount for it to decrease. And people on the road could be offered money to leave the road and park. People who will be going somewhere could be paid to delay their trip, but to avoid gaming that system you’d need to not pay them until they did make it to their stated destination.

The issue to figure out, then, would be how to get people off the road soon enough, and how to figure out the pricing on both sides.

Matthew C
Apr 6 2018 at 6:50pm

This was a great episode. As a longtime listener, I loved hearing you and Mike (undoubtedly my favorite guest due to the dynamic you two create) learn from each other in real-time.

Two topics that were touched upon- dead weight loss and Mike’s interest in the morality of voluntary actions- are both ideas I don’t often encounter, and would love if you could discuss them more in-depth on future podcasts.

Apr 6 2018 at 7:54pm

I thought this was a very interesting episode, but I didn’t quite follow some of the logic Russ and Michael threw out, so I put together a simple excel model.

It’s quite easy to show that the concept of externality is real in this situation (i.e. the road is over-used because people are making decisions based only on their own costs and benefits and not taking into account additional costs to everyone else). The fact that all road users impose costs on others is not a sufficient argument for ignoring its impact.

For example, in a simple scenario I build, I had 100 road users who each chose whether or not to use the road, each with varying benefits to taking a trip and using a logistic function based on the number of driver to determine congestion (a reasonable assumption – negligible congestion at first, then a steep increase, then a flattening as you approach the maximum amount of congestion). Using the scenario I built, 47 drivers would choose to drive based on their own utility function (assuming they can perfectly predict the amount of traffic – again, plausible with things like google maps), but the maximum total utility would be reached if only the top 35 people chose to drive, and the increase in output of the road (i.e. how much its use increases overall utility) would be about 40% higher if only 35 people drove!

Now, as Russ and Michael discussed, how exactly you go about charging users in order to discourage the people below the top 35 from driving is a very difficult practical question. However, assuming you could, you have successfully increased the output of the road, although you’ve actually decreased every driver’s utility.

From there, it’s a question of how wastefully the government spends money. In my particular example, if government waste results in the tax dollars being spent on 55% as efficiently as they would otherwise be spent, it’s breakeven. I think you could make a valid argument that by merely using these taxes to fund roads you’d actually achieve greater than 100% efficiency (because you could reduce taxes elsewhere and eliminate deadweight losses caused by them), though.

Tim Obst
Apr 7 2018 at 3:26am

Lots of great discussion. Like the host, I am not sure I understood all the points.

But for fun, instead of using the cliche story of looking for lost keys under the street light –because that is where you can see…

… Use the more original converse story of why nobody ever sees rhinoceroses hiding in trees — maybe because rhinoceroses are very good at it!

David Zetland
Apr 7 2018 at 5:03am

You guys are getting hung up on too many moving parts.

In Amsterdam, there’s less car congestion because the total cost of ownership and use are high. So many charges are involved.

For activities with negative externalites (contributing to congestion), raising the price REDUCES deadweight loss from overuse of the underpriced good. The use of the revenue is irrelevant, as usual, with respect to the negative externality, although it may be useful to subsidize non-congestion transport.

These mistakes regarding taxes are surprisingly common among economists who assume all taxes CAUSE dwl. More: http://www.aguanomics.com/2013/01/pigouvian-taxes-do-not-produce.html

Peter Pitsch
Apr 8 2018 at 9:56am

Listened to the podcast (huge fan) and read the thread. Here’s my two cents:

Imagine if the roads were operated by a private company. Assume for the moment, it didn’t act as a monopolist.

How would it price access when there was INEFFICIENT congestion? (Some reduction in average speed may be efficient if increases overall usage.)

Given today’s technology, it would ration this scarcity using the price system. This would enhance overall consumer benefit from the road which would now be used intensively in peak periods by those who valued it the most.

Distributional concerns should be addressed directly (give income to the poor) and not by distorting the price system.

In most cases, private ownership of roads won’t be practical or politically feasible. So the question is: can a government-run system approximate the above private company outcome without incurring offsetting costs?

Experimenting with such a system by starting with lower fees seems like a good bet. (We don’t want it acting like a monopolist.)

I believe the point about there being a lack of an exchange between consumers misses a fundamental point. The business in the above example, as is generally the case, would be acting as an intermediary for the owners of all the inputs that go into making the road available and the consumers who might use it at any one time. When I buy a car i don’t enter into exchanges with the millions of people who contributed to making and getting the car to me. Note that in the private company example above, those who refrain from using the road in the peak periods because of the fee would not be compensated.

The government’s use of the fees collected is a trickier question. One might say the fees collected must be reinvested in the production of road services or transportation services, but I don’t think that is the only right outcome. It might be the case that the managers of the “firm” in this case, that is the relevant government officials, might reasonably decide to return the money to taxpayers or do something else with the money that they believe the citizenry values more highly. If implemented at a local level, the democratic process could live and learn and vote accordingly.

Apr 10 2018 at 2:09am

Your widget un-analogy is inaccurate since no one person made the widget. We all payed for the roads, especially people who don’t use it at all (because they bike, work from home, work in the suburbs, or pay high rent close to work).

How did you not mention the tragedy of the commoners? (and Coas as someone already asked)

I agree it’s hard to estimate the right price of the road. currently it’s set at zero. How is that better?
Supposedly, the people priced out of using the road are compensated with faster roads for public transportation

The hope is that it is more efficient for (almost) everyone. For the rich people they get an empty road. The people who didn’t use the road anyway are better off because more tax is collected from other people (they no longer have to pay for roads, they get better public transportation…). The question is regarding people in the middle; priced out of using the road at peek hours, or begrudgingly pay the tax but would rather the traffic. Hopefully they will be better off (we hope) from the taxes collected from others, will have traffic free public lanes…

Still, remembering the externalities of using the roads (price, noise, congestion, pollution) I have to say I believe this to be a far better solution

Apr 10 2018 at 10:28am

Taking from Peter to give to Paul is wrong. It’s stealing.

But suppose you’re already taking $3 from Peter AND $7 from Paul every week and giving it all to Joseph. You decide to adjust these amounts for whatever reason, upping Peter to $4/week and lowering Paul to $6/week.

Does this count as taking from Peter and giving to Paul? This appears to be Robert’s position. After all, the net effect is a $1/week transfer from Peter to Paul. Yet, the new arrangement feels better than the original, so why not take a step in the right direction?

This is how I feel about applying a congestion tax. If the whole arrangement of taking money from Peter & Paul and giving it to Joseph is unjust (taxation), but the arrangement isn’t going anywhere soon, might as well choose the least unjust arrangement (using a congestion tax instead of an income tax).

Apr 10 2018 at 10:59am

The purpose of life is to dissipate energy gradients, currently fossil fuel energy gradients. Traffic does this pretty well, but congestion free traffic with a F-150 plus trailer going across open roads at 80 mph would be better.

To move toward this more efficient state it might be better to remove subsidies rather than increase taxes. More like the Netherlands mentioned above. Stop subsidizing population, automobiles, suburbia, and gasoline. Some of these subsidies are 1000 times the marginal value. For example, moving a 2000 pound car 30 miles with manpower alone at minimum wage would be 3000 dollars, with a gallon of gas, 3 dollars. Oil is more valuable than the pump price.

No, wait a minute, present system dissipates energy faster, may be close to maximal as it is.
Never mind.

Robert Wiblin
Apr 10 2018 at 4:44pm

It thought it was very odd when Russ said that charging for roads during congested times to free up money for other government services would be immoral because it would create winners or losers among the population (like any tax).

People who use roads are often subsidised by taxpayers at large who pay the fixed costs of making them.

Imagine that congestion fees were used to cover some of the cost of building the road network. Getting rid of those fees would then also create winners and losers (roughly the reverse of those created by imposing them in Russ’ example).

Is that also an immoral government redistribution?

In other words, why privilege a price of $0 and say deviation from that is bad because it creates winners and losers? That’s just an arbitrary price for receiving a scarce and expensive-to-provide service.

More generally, every change in government policy creates at least some winners and losers, but I don’t think that makes them immoral or forbidden.

Harvey Cody
Apr 11 2018 at 6:36pm

Great podcast!

A fair conclusion from the discussion is that there are no solutions to traffic congestion, and those “solutions” cities are using are not efficient and are far from optimal at achieving any objective. Absent from the interview are at least two points that confound the issues even more.

A significant focus in the conversation was on the “value of time” differential among drivers, and the tradeoffs of exploiting that variable to reduce congestion.

Two factors related to the value of time that were not mentioned are: 1) a person’s value of time varies based on circumstances, and 2) the value of money differs from driver to driver.

An example of the first point would be a retiree who plays Sudoku for most of each day who is nudged by a toll not to use the freeways during rush hours (say so that emergency room surgeons can spend more time at the hospital and less time in traffic). Not that a toll is necessarily justified, but that tradeoff has some societal merit despite the fact that the Sudoku devotee is not compensated as he would be in a market transaction. On the other hand, if the retiree’s wife cracks her skull on the floor and he needs to get her to the hospital immediately, then he is not a low “value of time” person—at least at that moment. A toll that is prohibitively high for him could make a life and death difference.

A surgeon on the other hand could be so wealthy that deciding whether to pay a toll to get to the beach for a family picnic a little earlier is not worth a moment’s consideration. The “value of money” is much lower for some people than others.

An upshot of the second point is that a congestion toll does, to some degree, represent a real cost to poor and a real benefit of the rich. That too has both positive and negative consequences.

As I discussed in “Tariffs Transfer Wealth From the Poor To The Rich,” tariffs have a similar effect. (See: https://harveycody.com/2017/01/27/tariffs-transfer-wealth-from-the-poor-to-the-rich/)

Apr 13 2018 at 1:46am

Is the solution to use the payments made by the road users to compensate the people who choose not to.

Issue everyone a right-to-use the existing roads, or sell rights-to-use to finance the construction of a new road.
Then set up an exchange where you can sell your right-to-use. People who don’t want to drive are compensated directly by those that do. Prices for popular times are higher.

Jim Thorson
Apr 13 2018 at 4:35pm

The comments on privatization got me thinking. A private owner of a roads incentive is to maximize the number of cars per hour. That is almost exactly in line with what we all want to do- get as many people on/ off the highway as possible.

To understand the right way to incentivize people to minimize traffic jams, we should understand both the physics/ mathematics of traffic jams, then use economics and/ or technology to get the correct behavior.
A simple traffic model shows that the number of cars per hour that can go thru a point on a highway is a monotonically increasing function of speed.( At higher speeds traffic is spaced out more but the higher speed of each car result inn a net of more cars/ hr) Thus, a lane can handle almost 2X as many cars at 65 mph than at 10 mph – a double benefit for commuters – more cars and higher speed.

This leads to the fundamental issue- the best action for all driver to take is to wait until the traffic is moving at or near the speed limit to enter the freeway. If ***everyone**** did did this, it would optimize the flow.

This is highly counter-intuitive. Waiting a few minutes to get on the freeways may seem like a long time, but if you have a 5 mile commute it will take you 5 minutes at 60 mph and 15 minutes at 20 mph. Thus you could “afford” a 10 minute wait.

In my area, we have metering lights during rush hour to regulate the flow onto the freeways. These unfortunately do not work as well as they could well for two reasons; 1. The are ignored by a significant number of people- (there is no risk of being caught, and most people don’t believe they are useful since they are counter-intuitive) 2. They cannot do anything about the traffic that enters before rush hour or from outside the area that is metered. (This leads to the belief that is an “unfairness” system, thus more people ignoring the lights. )

Technology could help here, since it is likely cheaper these days to implement these systems that it was 10 years ago. For example, we could monitor the length of the lines at every on ramp, and even have a checkpoint at entry points for outside thee city to control traffic density. By monitoring the length of the lines, we could ensure a first come/ first served system- if anyone would believe it worked and thus follow the rules.

Perhaps there a some other smart incentives we can implement to get people to wait?

The above discussion is about optimizing the capacity of the systems that we have in place. It also appears that the maximum highway capacity is 1500-2000 cars/ hr per lane, if the laod is well managed. This is driven by physics/ rational human behavior (self preservation to maintain a reasonable following distance), not economics. If demand is higher that this , we do need different incentives to shift the time of demand or build more capacity.

By the way, traffic is a great example of a self organizing system. As I said earlier, a lot of people ere skeptical of the freeway lights after they were installed. A senator was able to get a law passed to tun them off after they had been i place for a few years. It was a massive failure- commute increased by a large amount immediately, but after a few weeks commute time had dropped back to near normal levels. My interpretation is that it took a while for individuals to work out their best plan give the new traffic patterns.

Apr 18 2018 at 4:03am

Another thought provoking discussion, gentlemen.

A point not addressed is that these anti-congestion schemes can also have negative impacts on pedestrians. In the Orchard Road shopping district of Singapore, for example, pedestrians are forced underground at major intersections, or forced to cross two or three times instead of the direct one time in order to keep the traffic flowing. This attitude, that drivers are paying to use the streets and they not to be used for the convenience of pedestrians, also results in two story pedestrian bridges which must be negotiated just to cross a few lanes of road. In a wheelchair or on a bicycle? OK, use these time-consuming elevators put in just to keep you off the pavement. These barriers are off-putting, time consuming, and increasingly ignored by people in a hurry to get places without using cars and buses.

Apr 19 2018 at 7:53pm

In Australia’s large and fast growing cities the trend of the last 20 years has been to allow the private sector to build new arterial toll roads. In situations where no land is available due to existing buildings, tunnels are dug at extraordinary expense. Motorists who have a low value of time can still use the old routes and motorists who want to get somewhere more quickly will pay a significant toll of up to $5. We tend not to allowed time of day or congestion based premiums, which maybe reflects the political unacceptability of that approach? At any rate with populations in these cities doubling every 36 years this seems to be working.

Comments are closed.


EconTalk Extra, conversation starters for this podcast episode:

This week's guest:

This week's focus:

Additional ideas and people mentioned in this podcast episode:

A few more readings and background resources:

A few more EconTalk podcast episodes:



Podcast Episode Highlights

Intro. [Recording date: March 6, 2018.]

Russ Roberts: Our topic for today is traffic, or more accurately perhaps, traffic congestion, prompted by a recent article you wrote for Learn Liberty's website that we'll link to. And, our jumping off point is the idea that some people have proposed that Uber and other ride-sharing companies should have to pay a congestion fee in cities like New York, arguing that drivers spend time cruising around looking for riders and that slows everyone else down. This is what economists call a negative externality, and the argument is that by imposing a tax, Uber and its drivers, Lyft and its drivers, etc., will drive a little less; and that will reduce congestion, making people better off. What are your thoughts?

Michael Munger: It's such an interesting problem. Because, as you have often said, the phenomenon of traffic in the first place is interesting. Traffic is an emergent property of the desire of many people all to use the road at the same time to get to a place as fast as possible. And the result is, we all end up going really slowly and hating each other. And people really get angry in traffic. So, traffic is the collective and unintended but perfectly understandable phenomenon that we're all trying to use the road at the same time. The question is: Is there a way to allocate this scarce good more efficiently? And, one of the things that's interesting about the problem of congestion is that economists always think that there's an efficient solution to the problem. And their idea of efficiency is that we use the price mechanism to allocate it so that the last person to get on is basically indifferent between using the road and not using the road. And if you raised the price, it means that the people who don't value it very much, they'll wait; they'll do something else: they'll take mass transit. And the people who really, really need it are able to purchase what they want--which is a relatively fast trip. So, this notion was first introduced--the notion of using a price to solve the problem of traffic--was first introduced as far as I know by William Vickrey in a study that he did for the city of New York in 1952. And then, he published a series of papers on it, in 1968. One later won the Nobel Prize. The notion that many economists take about congestion is that it's a problem of matching: How can we make sure that the people who want this the most are able to use it? And, when I was thinking about this, this morning, it struck me that there's a problem of exchange. And there's a very deep point that you made on a previous EconTalk that I cite in the little paper in Learn Liberty that you said that you would put up. And, in it, it's that using price is not exactly the same thing as creating a market. And, the effect of taxing congestion. I mean, the problem with economists is that we often have this--it's almost a set of steps that we use automatically: If there's an externality, we should tax it; and if we have the right tax, the amount of the externality will be efficient and the problem is solved. You, actually, made an objection that it took me a long time to understand. I now think that I do understand it. And, remarkably, I think you were right.

Russ Roberts: Oh, my gosh. It's rare. Go ahead. Before you do, can I--although I want to dwell on that last point as long as possible; and I hate to interrupt you when you are in such an eloquent and profound set of insights about my insights.

Michael Munger: Yeah. Just because I said you were right. I understand.

Russ Roberts: Enough about me. Let's talk about why you think I'm right.


Russ Roberts: No, I want to back up just a tidge, and remind listeners that the word 'efficiency' when used by economists is not the same as its use in the everyday English language. In everyday English language, it means sort of, I would say, effective. Or, at lowest cost. There's a piece of that in the economics definition. But, when economists say 'efficient,' what they usually have in mind is the idea that the pie of economic wellbeing, or benefit, or benefit above-and-beyond cost--what we would call net benefit--that that net benefit could be bigger if we put a congestion tax on. That's what we mean by efficient: typically, that the net gain across all people in the country, or even in the world, is--grows. And your point about--I actually, I think I only have two thoughts on congestion, one of which you've made--which is that prices to not a market make. Markets have prices. But, imposing prices don't necessarily re-create what a market does. It can, in theory. But, doing that from the top down as an economics engineer is a little bit tricky. My second point is more about the distributional impact of that. But I want to make the--I want to actually go back to something you said earlier, which was: You said, 'There's a problem. There's too much use of the road, and the price then will limit the number of people on the road.' Another way to look at the problem is that the road is un-owned. It's effectively a commons. It's owned by all of us. So, of course, as you point out in the article, it is rationed. It is rationed by a price. It's just not a money price provided by a seller, or a group of sellers. It's the time price. So, when it's not rush hour, at these points the conversation is not relevant, or when there's not traffic. But when we're talking about traffic, which will be our presumption throughout most of this conversation, the fundamental problem is that the driver getting on the road doesn't take into account the fact that the other drivers will be slowed down. And, as a result, too many people then enter the road. And, as a result, there's a potential gain from reducing the number of people on the road. And my point, which is, in many ways it grows out of a point you've made, many times, about the challenge of choosing in groups: My point is that it's not so easy to fix that problem. You may think it is, but it's not so easy.

Michael Munger: It isn't. And as I said, thinking about it, I want to emphasize both of the points that you just made. One is the information problem; and the other is the distribution problem. So, this, it may seem a little bit tedious to take a sort of deep dive into the philosophy of voluntary exchange; but it's something that I'm really interested in. So, if you'll indulge me for a minute or two--

Russ Roberts: Take your time.

Michael Munger: Well, for a truly voluntary exchange--one of the reasons that economists are so interested in truly voluntary exchanges, it means that both parties to the exchange, if it's not coerced, are made better off by the exchange. Which means that the voluntary part of that also has ethical consequences. So, philosophers think of that as being an erogatory obligation. Which means that in the normal course of things, I have--it behooves me to behave that way. And, in commercial activity it means--an erogatory obligation means--I am not obliged to harm myself. Now, a super-erogatory obligation is charity or the Good Samaritan. So, you see that I'm drowning; there's a life-ring right beside you with a rope; you have a good strong arm. It would be pretty bad of you not to throw it. But you are not obliged to get into the water to try to save me. So, you are not obliged to harm yourself.

Russ Roberts: Although, it would be nice, if I sacrificed my time and put myself at risk.

Michael Munger: Yeah. It's a different thing, where we would say, 'That person is behaving in a way that's very praiseworthy, because the potential of sacrifice--I've given up something to try to help someone else--many of us would say, that's: You are both loved and lovely, if you behave that way. But it's a different context than commercial activity, which is just erogatory. Now, let's think about the way erogatory exchange works. I have a widget that I value $1. You value that same widget at $5. Who should own the widget? Well, you could argue that you should own the widget. And, the nice thing about markets is that it provides with two pieces, two parts of a process. One is, the price tells us that you value it more. Because, you'll pay $5 and I'll take $1. The other thing is the distributional consequences are solved, because you and I can negotiate a mutually beneficial exchange of, let's say, $2.50. You pay $2.50, you're better off, because you'd pay up to $5. I get $2.50 and I'm better off, because I would take anything over $1. That means that there's an actual exchange; the world is a better place. Because the widget moved to a higher-valued use. Now, the question is: Can we use, can we--can a central planner, comes in, and notices this, and says, 'Well, I can match that. All I need to do is make sure that resources are moved to higher-valued uses.' Problem is, that the central planner lacks both parts of the process that we just talked about. Doesn't have price. Has to make a guess about how much I value it, and how much you value it. And, since what is being taken from me and given to you is not compensated--the distributional consequences are potentially a problem. So, when it comes to congestion pricing, I, the person that don't--the value of my time is not very high. I could sit home. I could wait. Suppose I'm going to go to the store: I'm thinking of going to the park. If the road is pretty busy, I might wait. But, I don't mind waiting in traffic. The time I might value--the value of my time is not very high. So, I go ahead and I get out there on the road. The problem is, that contributes to congestion. The estimates of, you know, the amount of congestion that an additional car causes, are between 3 and 10 times the amount of the cost that other cars impose on me. So, there's a complicated process that uses simulations. But, I could just in time[?], adding another car seems to impose higher costs on others that are those that are fully borne by the marginal car. I don't mind, though, because the opportunity cost of my time is very low; and so I go ahead and get out there. You, have to go to a doctor's appointment. Or, you have to go to give a lecture where you are going to create a lot of value and get paid quite a bit of money.

Russ Roberts: Love your optimism--

Michael Munger: Hah, hah, hah. I said, 'value,' too. I understand wealth.

Russ Roberts: Keep going.

Michael Munger: I'm going to create a lot of value. But, there's a thousand people that are waiting for you, Russ.

Russ Roberts: Uh, hoh, hoh, hoh.

Michael Munger: And they are looking at their watch; and they are saying, 'Russ isn't here.'

Russ Roberts: And their breath is baited. I don't know what that means, but I'm sure they are sitting there with baited breath.

Michael Munger: It's because of traffic. They all think it's that darn traffic. You would happily pay for some people to wait. And there are people who, if you could pay them, would be willing to wait. But there's no way to effect that transaction. The mistake is to think that we can somehow approximate that with a congestion price. We don't actually know that the people who are going to pay this are the ones who have the highest opportunity cost value of time. The reason is--and this is kind of complicated but it's the reason I went through all the other stuff--the reason is: the last car to get on the road is the one that pays the congestion tax. But, the marginal user is the one who happens to value the time least. Now, that person may already be on the road--


Russ Roberts: Time out. Time out. I don't think that's right. I apologize for interrupting, but I don't want to go down what doesn't seem right to me. Or maybe I misunderstood you. A congestion tax is imposed on all drivers, at a particular time. Not one--

Michael Munger: Not a dynamic one. A dynamic congestion tax is only imposed once the congestion is bad.

Russ Roberts: Right. Sorry. Sure. So, at 5 p.m.--let's make it simple. Starting at 5 p.m., all cars pay a dollar a mile premium. And, we're going to avoid the toll-collecting problem, which we can, now, through technology: we can all have E-Zpass, a fairly cheap, minimal expense way to figure out who is on the road, when. We're going to ignore the privacy issues that that raises. And, we're going to say that we can monitor when people are on the road and for how long. So, in my city, Washington, D.C., we have such a system on certain roads at certain times. It's not all roads. And it could change every day--it could be literally dynamic, dynamically dynamic. But, I want to just assume that everybody knows that starting at 5 o'clock, there's a surcharge of a dollar a mile on a particular road.

Michael Munger: That's not what I was talking about.

Russ Roberts: Yeah. I don't get your example. I've never heard anyone propose your example. What's your example?

Michael Munger: My example is actual dynamic pricing. So, let's go all the way to dynamic pricing. So, there's a bunch of people on the road already. And, many of them have a low opportunity cost of time. Now, we charge anyone who enters--these are entry fees, and there's a lot of cities that use these--for you to get on the Interstate, you have to pay an entry fee because it's already congested. But you may be a high-valued user. All the cars that are already on there are low-valued users. That's what's causing the congestion. So, the point that I'm making is actually quite a simple one. The question is: At what margin can we charge this? There's no efficient way to charge the low opportunity cost people if you have a dynamic pricing model, which people take to be the gold standard of congestion pricing. Now, what you said is, 'We'll do it every day at 5 o'clock.' Most economists would say that's second best; but it may solve the problem, because, as you say, I know that at 5 o'clock, I'll have to pay for it, and if I'm a low-valued user, I'll avoid the road at that time. Because, then, it's an average, not a marginal cost problem. The pure dynamic one, though, the dynamic pricing means that the road's already congested, because there was a zero price up until the point it was congested. Many of the people causing the congestion are low-valued users.

Russ Roberts: But I don't understand that. I mean, I literally don't. If we only charge a fee for when the road is congested, that encourages people to get on the road before, and start congesting the road earlier. There is no difference between me, who got on the road at 4:57 with thousands of other people--and suddenly it just gets to the point, it piles up to the point where traffic starts to slow--its average speed starts to slow down. And then we say, 'Okay, well now if someone gets on, we're going to charge them.' But I'm just as responsible for the externality as the person who is getting on now. I'm on the road, slowing it down, just like that person getting on the road now will slow it down. It doesn't seem right to me. I think you have to charge all cars, on efficiency grounds, which I'm going to fight against. But to accept the argument of the standard, what I think is the standard argument by economists on road congestion is, there should be either--I mean, it doesn't help, to put a tax--as some cities have suggested, there's going to be, and I think some cities have done this--a congestion fee if you enter the central city at a certain time. After a certain time. Well, what about the people who have already entered? They get off for free? That just encourages people to get on earlier.

Michael Munger: But that's my argument. That's what they're doing.

Russ Roberts: Okay. Explain that.

Michael Munger: Well, you just made the argument. You do understand it.

Russ Roberts: I don't think so. I'm so smart. I can't every understand my own argument. That's how smart I am: My argument is so clever, I don't even understand it.

Michael Munger: I literally [?]

Russ Roberts: I think you've understood your version of my argument. Which is interesting. Just not mine. But, go ahead.

Michael Munger: I am literally going to repeat what you just said. There's a bunch of people already in the central city. That's what's causing the congestion. We're going to trigger congestion pricing once there's already congestion. That means that some of the people already there who didn't have to pay, have a low opportunity cost of time. So, you're saying--

Russ Roberts: Some might have a high opportunity cost of time. It's mixed.

Michael Munger: Yes. But even the ones that have a low opportunity cost of time, it's already congested. Because we didn't charge them. And so, you're saying you don't like that pricing model. I don't disagree with you. We agree about that. But that pricing model means that we are working at the wrong margin. So, the margin that we want to do is to somehow enable the high-valued users ideally to compensate the low-valued users. To say, 'It would be better if you would wait.' So, in the case of the widget, I'm going to pay you. You actually receive it. I get it; we know that the prices work; it's a voluntary exchange. The question is: Is there a dynamic congestion pricing model that approximates that?

Russ Roberts: Yeah; and the one you gave certainly doesn't. Well, now I think I understand your point. You are saying that: if I charge a fee after the point at which the roads are congested, there are a lot of people on the highway already who, people who aren't on the highway would be thrilled to pay, leave and take their spot. Let's make this clearer. Let's say that past 10,000 cars on this particular road, it starts to get congested. So, as soon as it hits 10,000 and starts to climb above it, they put the congestion fee on. Some of the 10,000 are already there are going to be people that would be happy to bribed to leave, let someone else be one of the lucky 10,000. And as a result of this congestion fee, that's not going to work. That congestion fee will not satisfy that requirement. That, to me, is not the problem. It's interesting. It's fascinating that that's what you've learned from our earlier conversation.

Michael Munger: Well, no; that's actually what I thought about this morning; and it turns out that that's a problem that many people in this field worry about--is that, there's a kind of market failure, because the people who want to buy less congestion can't pay those who would be willing to sell less congestion.

Russ Roberts: Yeah, but that's--yeah. Sorry.

Michael Munger: And it's all masked behind the fact that it's apparently a pretty reasonable thing to have a congestion tax based on: 'The road's congested; anybody who also shows up now has to pay an additional entry fee.' That just turns out to be a really bad idea.

Russ Roberts: Oh, that's for sure. So, let's change it. Let's pick--instead of the next, everyone else who shows up after this time, let's say everyone who is on the road at this time. And that solves the problem. On the surface.

Michael Munger: Yes.

Russ Roberts: I still think it's wrong. I'm going to argue why it's wrong. But that's clearly a better tax than just 'Everybody who gets in before the time, before the congestion starts is somehow absolved from having to pay the tax,' because that just encourages them to get there a little bit earlier.

Michael Munger: So, let the record show you went from saying, 'I'm completely incorrect to obviously right.'

Russ Roberts: Fair enough.

Michael Munger: And that it's obviously right in a way that's boring. But I wanted to start with baby steps. There are cities that use this. That's clearly wrong. But is there a better model?


Russ Roberts: Time out. I wouldn't say 'It's clearly wrong.' To me, I would say, 'It's clearly not efficient,' in the sense that it doesn't not guarantee that the highest-value people use the road.

Michael Munger: It just doesn't even accomplish its own object. It's own objective is to try to match up--it's working at the wrong margin. So, now we should go to more interesting pricing models. I agree. I'm sorry I belabored that.

Russ Roberts: That's all right. It's interesting. Go ahead.

Michael Munger: Well, so then, let's suppose we have a better pricing model; and in it we will--one way to do it is to start charging every day for higher prices when normally congestion hits. And that means that the people who have a low opportunity cost of their time knowing that they'll put off trips, they'll take mass transit, they'll do something else; and the people with the high opportunity cost of time will still pay it. And there's some hybrid versions of that. There's HOV [High Occupancy Vehicle] lanes. So, you can have just one lane on the highway. Now, there, HOV, since it stands for High Occupancy Vehicle, you have to pay on a different margin. You have to get other people to ride with you. But you can also have--it would be simple enough just to charge a higher price for one lane. So, you could have a pricing model; and in Santiago, Chile, the Costanera Norte is a private road effectively that goes through the downtown. They charge pretty high prices all the time. There's alternative roads, but the Costanera Norte, almost always there's very little congestion. And so, if you just pay extra, you can buy a lack of congestion. You can buy a faster trip.

Russ Roberts: And I want to just--it's important to mention--you keep saying low and high values of time. Of course, it's the net value of the trip relative to the time that people are taking into account. So, it's not just that the people with the highest value of time are the ones in the toll lane or who would go when the congestion tax is set. It's the people whose net value--the value of the trip itself minus their opportunity cost of time, what they would be doing with their time otherwise--who go. Right? It's not just allocating according to the value of their time.

Michael Munger: Yes. The interesting thing about that, is that one of the reasons that--because this is the money value of my time. And, one of the problems is that our mythical planner who came in a thought 'I can use a tax system' in effect to approximate the good results of voluntary exchange, is that the sort of utilitarian consequences of this will be positive. Because, it's as if I bought this--I who value this a lot bought it from someone who valued it much less--in money terms. That was the reason I was talking about the opportunity cost of time: is that, there's a sort of mythology that in effect there's a kind of transaction that's taking place. But, there's not. This is something that's only taking place inside the mind of the planner, who is using a kind of utilitarianism: That is, the people who have low opportunity cost of time, it's okay that they sit home. And that really was the heart of your objection in the podcast a few years ago, the Winston podcast a few years ago. You can't know that. Because, there's a couple of deadweight losses. One deadweight loss is the deadweight loss--

Russ Roberts: Explain what deadweight loss is.

Michael Munger: Deadweight loss is a loss that is just dissipated. It's not a transfer. So, if I have to pay you something, I lose but you gain. If I just sit in traffic, or I wait in line, then the value of my time is just burned off into the atmosphere. So, I could be producing something; I could be making something for society. I could be enjoying maybe going for a walk in a park. But, some value is foregone. If I sit for an hour in traffic, I burn up an hour of my time and no one receives the benefit. And so economists call that a deadweight loss.

Russ Roberts: And for listeners who might have the following thought is really important: You might think, well, you're not going to waste your time in traffic. You're going to be listening to EconTalk. And what that means, though, is that the amount of traffic there's going to--if everyone can listen to EconTalk--unimaginably, but it does happen, other podcasts; or read, listen to a book on tape, or listen to some other podcast--then the time cost of travel goes down. And therefore more people will be willing to get on the road. Because it's not as unpleasant. Which means it has to be a long enough trip to make it, even with EconTalk, somewhat unpleasant. That's what congestion does. Congestion rations access via the fact that you burned time. And you say, 'Well, you won't waste it. You'll shave in the car, and you'll put on your makeup. And you'll listen to EconTalk or books on tape.' That just means that more and more people are going to get on the road. Because, otherwise the road's underpriced through time. It's not being allocated. People will find it--I didn't say that very well. If it's pleasant to travel, that just means more people will get on the road. And at times of congestion, more and more people would like to get on the road than there is space available. The opportunity to do things that make the time go by actually just means that you are going to be on the road longer.

Michael Munger: Yep, because the disutility--how much I dislike being on the road is reduced. Knowing that I'm going to be on the road, I'll plan for that. And there certainly are a lot of things--cars are more comfortable; the control that we have of the air inside that is better. It really is true that longer commuting times, we're likely to come up with ways to make that less painful. Which reduces, it's true, the deadweight loss. But the point that you made--

Russ Roberts: No, it doesn't reduce the deadweight loss. That's my point. It has to persist. Because otherwise there's no allocation going on. Let me give a goofy example. The example I use in class is that I'm going to give an A+ to the first 5 students--I'm going to give all the answers out to my exam: let's say I have an exam coming up on Friday; and on Thursday night at midnight, I'm going to give out the answers to the exam to the first 5 students who are outside my door. And--this is a stupid example; it's immoral; forget all of that for the moment--but the idea is that there's some group of students in the class who are worried they are going to get a bad grade, who don't have time to study. There's a lot of motivations. And they're going to show up at midnight to get those answers and get a guaranteed A+. That's the goal. So, if I do that--and let's say there's 500 students in my class--if I do that, how many people are going to wait in line? And the answer is 5, because the 6th person realizes they can't get anything. There's only room for 5 people. That's the deal. That's my offer. So, what happens? Well, you have to get there early enough to be one of the first five. So, when would that be? Well, the first year, who knows when that will be. But, let's say after a while it becomes known that if you get there by 2 o'clock on Thursday even though the thing is not given out till midnight, you'll be one of the first 5; and that turns out to be the general pattern. And then after a year, after a while I feel bad. After a few years I say, 'You know, it's terrible. These kids have to sit outside the room for 10 hours with nothing to do. It's just cruel. I'm going to put up some really nice armchairs. And, they'll give them a massage while they're sitting there. And they'll be really comfortable. I'll put a drink holder in. And I'll show movies. So, that way they won't waste their time for 10 hours while they're sitting there. And if I do that, they'll just get there earlier. It won't have any effect on the deadweight loss. They'll just get there early enough now to be one of the first 5 when it's relatively pleasant. Which is before 2 o'clock. Now they may have to get there the night before, or two nights before, to be one of the first 5. So, I don't think that affects the deadweight loss.

Michael Munger: Well, you're right; but you just skipped over the intervening step. Knowing I'm going to--I was trying to do it in two steps--knowing I'm going to wait in traffic, I'm going to get a more comfortable car; I'm going to have a phone that's full of EconTalk, and yes, it's probably true--other podcasts as well.

Russ Roberts: Hissss.

Michael Munger: So, the result of that is that I'm then willing to wait even longer in traffic. I'm less concerned about traffic. And, it ought to be at the same margin--the deadweight loss ought to be the same. It's just that the cost to me per minute of being in traffic is reduced by my anticipation of this. But, overall, yes: the deadweight loss will be the same. It just takes to get there. What I think we're ignoring--what I think some of this consideration, which is what you had picked up in the earlier podcast--if someone wants to go to the store, wants to go to the park, needs to get to work, but the congestion pricing prevents them from doing that, that's a deadweight loss, too, because those are trips that are not taken. That is, I don't go to the store. I stay at home. I sit at home instead of doing the things that I want to do, because now the price has gone up. I'm not compensated for the fact that something useful is being taken from me. So, the--you can say it's a low-valued user, or maybe it's just a poor person who can't afford to pay the price. In a market that works out to be the same thing. They don't objectively in terms of utility have a low opportunity cost of time. They just can't afford to pay the congestion tax. The result is that deadweight losses are imposed on them. And, the hard thing is to balance the deadweight losses of sitting in traffic compared to trips not taken, lives not lived, because some people can't afford to pay the congestion tax. That doesn't mean it's a terrible idea. But it's not the same as a market. So, in a way the point you were making is quite simple and it just never occurred to me before. And I wanted to give you credit for it.


Russ Roberts: I have no idea what you're talking about. So interesting. I don't know if anyone's still listening to us, Mike. I'm always happy to talk to you even if we're just on the phone chitchatting. And I don't know if we've lost all of our listeners. But, whether there are others who have just raced to their phones to hear this because it's so interesting. I'm hoping it's the latter. But, I'm fascinated by the fact that that is not at all what I meant. This is a different point than we got tangentially discussing earlier. I'm going to try to make what I consider my point; and then we're going to see how it interacts with your point.

Michael Munger: Actually, this morning when I was preparing for this, I said, 'I fully expect an Annie Hall moment,' where I say, 'Well, this is what Russ Roberts meant'; and Woody Allen pulls out Marshall McLuhan and says, 'No. Here's what he really meant. You're an idiot.'

Russ Roberts: At least it's not at the point where my payos are growing at the dinner table. But--that's just a Yiddish reference, for those of you keeping track at home. So, here's the way I think about this. And I love the idea of the exchange part of this. And, I wrote a piece a long time ago about--we're going to come full circle to all the past Mungerisms of past EconTalk episodes. You and I have spent, I think, at least two episodes talking about price gouging. And, my point about price gouging is that--one of my points, is that when prices rise, the example I give is that I was going to build a porch on my house in St. Louis. And, my architect, the guy who built up the plans said it's going to cost--whatever he said; I think we were trying to spend, I'd say $10,000. And then the bids all came in at like $30,000. And I was kind of annoyed at the architect. I felt that he had taken advantage of me. He had encouraged me to hire him to draw up some plans for a project that I had said I would do in the $10,000-ish range. But now he was off by a factor of 3. And then I realized, being a slow-witted economist but an economist nevertheless, that the reason it was $30 instead of $10 is that there had been a flood. The Mississippi had flooded. And building a porch was not nearly as valuable to most people as putting back a wall that had fallen down. And so, carpenters charged a premium. And as a result of that, I said, 'You know, I'm going to wait to do that porch. I'm not going to do that, actually.' And when I realized it was because of the storm, I thought, 'Well, I'll try again in 18 months.' Or it was 12, or whatever it was. Or 18. It was something more reasonable. To me. But for somebody whose house was falling down, they were thrilled to pay a premium to get that wall put up by a carpenter. So, what the price does in that setting is it encourages me to step aside. Basically, the person whose wall has fallen down pays me--not pays me, but I am induced to forego using the carpenter and let the carpenter go to the higher-valued use of the person whose house is falling down rather than just adding a porch--which is pleasant but not very important relative to a house. We agree on that, right? We're good?

Michael Munger: Yes. Sure.


Russ Roberts: Now. I didn't think of this the way you are thinking of it, in the context of congestion. The way I think about it is as follows. And it's nothing like the way you think about it. Even though you "learned it from me." But, I think I know why--I think I see why we diverge. So, here's my point about the congestion pricing. I like to start with the world where everyone has the same value of time. But, we differ by, say, the value of the trip we are taking. Okay? So, what's going to happen when we put a tax on, is that we are going to make the road "more efficient." It's going to mean that the time spent is going to be less. Which is great. But my claim is, is that in that world--and this is very unintuitive to most people--in that world, most economists, in that world every driver is worse off. Every driver. Every driver. Plus the people at home. The people at home are worse off because they don't take the trip now. The people who are on the road are worse off because they are paying a tax. And you might say, 'Oh, but they save time.' And my point about the tax is that it has to be the case in some sense that the pain from the tax has to outweigh the gain from the savings in travel time. That is, you might think, 'Well, who knows if you are better off or worse off if you take the trip? In fact, you are probably better off: You took the trip.' And obviously you got there quicker than you would have. That's the purpose of the tax. It took out the congestion. And so it's true you had to pay the tax, but you got there quicker so you are better off. And that's false! That's basically false. That's basically why I'm hand-waving there: maybe not worth explaining. But we'll see. But my point is, is that the whole point of the tax is to discourage people from getting on the road. And therefore, it has to be the case that the pain from the tax paid has to be larger than the gains from the fact that it takes a shorter time. Then you say, 'Well, whoa, that's not efficient.' And the answer is, 'Yes it is.' Because--that, you could argue, 'Well, you didn't set the tax right.' And I would say, 'You exactly--that has to be the case, under a correctly set tax.' Because the efficiency gain, the net gain to all of society comes from the fact that the money collected from the tax can go for some other purpose. So, when we look across society after the tax is put on, drivers are worse off. But the people who get the benefits of the money are better off. And that gain has to be larger. And that's true. You can show that with a graph. But it doesn't accrue to the drivers. It accrues to "someone else." And then people say, 'Well, let's just give it back to the drivers, then.' And the answer is: Once you do that, you are going to start to undo the incentive effects of the tax. 'Oh, we'll do it in a lump-sum fashion so it doesn't affect incentives.' Well, it's really hard to do that. It's almost impossible to do that. And it's really hard to do it with any accuracy. Now: in the real world, of course, people have different values of time. So there will be drivers who are better off. The people who have very, very, high values of time, or higher than average values of time, or higher than median values of time, depending on assumptions you'd make--those people are going to be better off, because for them, that small group, their value of time is so high that the tax is not going to be large enough to offset the gains from the less-congested road. But for everybody else--they're just worse--they are punished by this law. And they don't get compensated for it. So, what you've done, in my view, with the congestion tax, is you've made "society better off" in a utilitarian way. If you added up all the gains and losses, the gains would outweigh the losses. But, you'd have just arbitrarily punished a bunch of people to create this net gain for some others who happen to get the benefit of what the government spends the money on. And that just seems to me to be immoral.

Michael Munger: Maybe not any more immoral than most taxes. But, no less, is your point. It's just a tax. And the other argle-bargle about benefits and distribution isn't right. It's really just a tax, and we should think of it that way. And, if that's the best way of raising revenue, well, okay. But it depends what you do with the money. Usually the argument, as you know, is we'll use the money raised from the congestion tax to subsidize or expand mass transit. Which should benefit those who now no longer can afford to use the roads. So, we have better buses; we have better subways. I think that would be the answer: that, it does matter what you do with the money. If what you do with the money is--

Russ Roberts: I'm thinking food stamps. Expand the food stamp program, or build better schools. Or--pay old people. Higher Social Security benefit. Old people would benefit. It's like: Who would ever imagine advocating for the following: 'You know, there are a lot of people with low value of time. Let's tax them to pay for Social Security benefits.' You'd say: Well, that seems wrong. That's almost as wrong as--I don't know, say, using lotteries to fund schools? We're going to take people who are desperate for money, who probably dropped out of school, and use their pitiful money that they are spending on their lottery tickets to help people who--it just seems bizarre. But that's, I think what we're doing with roads. And I think we're under the illusion, as economists--that we say, 'Oh, it's efficient, 'and that means everybody is better off. No. It means everybody could be made better off. But in reality, they won't be. So, is it moral, then, to impose a congestion tax?

Michael Munger: Well, it is an interesting question: What should be the price of roads? And, most--the efficiency condition is usually price is equal to marginal cost. If there's no congestion, marginal cost is close to zero. Because, I--maybe there's some wear and tear on the road. Big trucks probably impose some of that on the road. But cars, largely the marginal cost is zero. The problem is, with congestion you say that there is something--that I'm imposing some cost. It's not clear that we could figure what that price is. Your point about the tax is an interesting one. The thing that the article took up was congestion pricing on the island of Manhattan; and in particular, I actually got some response to this. I solicited some answers from readers. And the question is: Suppose that we agree that there's a problem with congestion on the lower part of Manhattan. Should--is some kind of congestion tax an answer? And of course, what some people in Manhattan say, particularly taxi drivers, the yellow taxi drivers, what they want to do is impose a congestion tax on Uber and other ride-sharing services. They say, 'At the margin, what's causing this is all these ride-sharing cars.' And, when you look--well, there's 60,000 Uber drivers that sometimes drive in Manhattan. And there's probably between 15,000 and 25,000 most days. And that's compared to 13,500 or so yellow taxis. So, let's go to a third pricing model. We've talked about two: one, just the marginal drivers; one, have a congestion tax generally. Should we tax some groups who, at the margin, have less of a right to use the roads? Because what the taxi drivers are claiming is that the ride-sharing services have less of a right than the taxi drivers; and they ought to have to pay a tax to offset that.


Russ Roberts: Yeah. My first thought, of course, is that I don't think traffic in New York started in 2009--when ride-sharing started. Has it? I don't know.

Michael Munger: It has gotten worse.

Russ Roberts: It's an empirical question, right?

Michael Munger: Yeah. Right. But empirically, it has gotten worse. Now, one of the big problems is double parking. So, even a little bit of double parking at rush hour is a big problem. And a lot of the ride-sharing--because if I drive around constantly, I'm using up gas. There's no place to park. And so, it is true that the big problem that you have is there's a delivery truck at 4:30 on one of the big avenues. Or, an Uber car is double parked or parked in a way that slows down traffic. Presumably you could just enforce the parking laws better and prevent double parking.

Russ Roberts: It's not the [?]

Michael Munger: Well, what the taxi companies want to do is impose a differential congestion tax. And what that made me think about was the sort of history of the taxi industry. And I don't know how deep you want me to go into that. But, the taxi industry in New York has a very interesting history. And the fact that they use these medallions was because, I think most of the listeners probably know, in order to drive a taxi lower Manhattan--that is, one of the yellow taxis, not one of the green borough taxis. But, a yellow taxi, you have to have a medallion. And medallions, the price for a while was up over $700,000 for one medallion. They were established in 1937 because there were so many taxicabs; and so many people were trying to drive so many hours a day during the Depression that--partly for congestion reasons, but partly to make sure that the taxis could make, they would say, a decent living, they restricted the number of taxis that could drive legally. So, this is the legacy of a time from during the Depression when prices were quite low. The result was the number of--I think the number of medallions was 16,000. Medallions were not even valuable enough to pay the $10 annual fee to renew them. So, about 3000 people just let their medallions lapse. Now, we're down to about 13,000. They had a couple of auctions where the value was pretty high. But now medallions are extremely valuable. And the price of a taxi in New York is not that high. The density of use, if you compare it to other cities, taxi rates in New York City, you are not unreasonably high. Why not just use the medallion system? Why allow Uber at all, is the question that the taxi drivers would ask? 'You are saying that we have a problem with congestion. We paid for our medallions. Isn't the solution just to outlaw anything that doesn't have the medallion?' Isn't that the solution? Forget this congestion tax. What we need instead is this third model of better regulation.

Russ Roberts: I mean--it's--well, there's no reason to think that 13,600 is the right number of medallions to have in the city. It's, as you would be the first to point out, it's the result of a political process, not an economist trying to figure out what the right number is. Obviously, the taxi cab drivers like it this way. It's easy for them to find rides and to fill those rides at the prices that the taxi commissioner sets, which, as you say, they're not unreasonable. But I think Uber is cheaper.

Michael Munger: Well, Uber must be cheaper, net. The time spent waiting, the inconvenience of giving directions, the inconvenience of paying, or otherwise people wouldn't use Uber so much. So, they really do use Uber a lot.

Russ Roberts: And I raised the question, the empirical question, of whether traffic has gotten worse, because it's--there are two thoughts I have, which I hadn't thought of before. One is: It's possible fewer people are driving around in their own cars because they don't have to have them any more. They are using Uber. They're not parking them on the street; there might be more parking available, because people don't use cars as much because they can trust Uber and it's relatively cheap. So, you would think the access to ride-sharing would reduce the number of cars, at least that are owned in New York; it doesn't necessarily reduce the number of miles because now with the lower price you might drive more. It's complicated. Of course. The second point is: I wonder if Uber has an incentive, and Lyft, to take account of congestion when they set their prices. Because, surge pricing--we think of surge pricing as a--okay, demand's higher because it's raining or it's rush hour or it's after a concert has gotten out in some particular part of town, and so there's surge pricing because there aren't enough people willing to drive relative to the increased demand. And Uber then raises the price that drivers can get and charge. And they get a share of that. So that helps ration the shortage. But it could be the case that--wouldn't they not want to slow down their own drivers? Which--I mean, obviously they don't have a monopoly; they don't own the roads. They don't have a full incentive to internalize that externality. But you'd think it might have some.

Michael Munger: There are a lot of mysteries in Uber's algorithm for deciding price, and I don't think anyone--it's proprietary. No one's announced what it is. Perhaps--well, like a taxi, if you sit in traffic for a long time without moving, you are also charged for that. So, that is part of the algorithm on Uber. But, that means that I'm more likely as a passenger, as we've said before, to try to find some other way of riding. Including walking, if necessary, or taking the subway. So, if the surface roads are really, really congested, I may try to take something else. It's interesting that the origin of taxis in New York--I actually had wondered why it was that the taxi industry developed the way that it did. You may know this story. But, in 1907, apparently a man took a taxi in which he was disgusted with the price. So, apparently it was about 3/4 of a mile. He took a taxi 3/4 of a mile and was charged $5. Now, I tried to investigate how expensive $5 really was, and it turns out that a full meal at Delmonico's, one of the most expensive restaurants in New York, would have been about $2.50. So, it's the equivalent of $125. He was charged $125 in 2018-dollars for a Hansom cab. And he decided that he would start a taxi company. Almost immediately, he started having labor troubles. So, the regulation of taxis in New York has a long and contentious history. The taxi drivers--I actually wonder if it's something like, in terms of industry structure, if it's something like monopolistic competition. And so that actually raises the question you just did, about Uber and its pricing model. Uber isn't a monopoly, but they do have a monolithic pricing model. They should be able to--

Russ Roberts: They have a large impact on the condition, the roads--

Michael Munger: And it's dynamic[?]. They can update it. They can update their pricing. So, monopolistic competition, what happens is, you get entry, where I don't serve much but locationally I can serve the people that are around me. And monopolistic competition is something that was made up first by Joan Robinson at Cambridge University. And restaurants or other things that have local monopolies work that way. I had wondered, and it's interesting that you brought that up about Uber's pricing model: Don't they have some incentive for solving the congestion problem? And they might--taxis, the problem with taxis is they are locked into a pricing structure. They can't change their price. It's always the same. Which means that sometimes the price is too high, and people who would pay more than the cost of giving a ride still take the subway because the taxi can't cut its price. And sometimes, during a rainy night or after the plays let out, or a sports event, a lot of people would pay more. But they can't do that. So, taxis can't adjust dynamically. Uber has, in a way, a better pricing model. That was all just a long way of wondering, if you're not right, might Uber account for the fact that there's an entry problem that they want to worry about. And congestion harms everybody. Why don't they charge a little bit higher price? I just don't know--I haven't seen any evidence that that's true.

Russ Roberts: I mean, the point being that if you had private roads, and it was easy to start a new road--it's not, obviously. But if it was easy to start new roads, and--I mean, if you think about flying. If you think about putting roads on top of each other--this is like the greatest EconTalk ever, Mike. It's got more--let's assume a can opener. It's got so many of those. But they are actually good ones, in my view. But that's what an economist would say when they make a ridiculous assumption. So let's, if we had different roads and we could just build a road on top of one another, it goes from Point A to Point B, but doesn't interfere--which of course you can't do. But if you could, and then you let people do that and enter, then there'd be privately held roads. And none of those would have congestion. The point being that, when you wait in line at the supermarket, in general, as you point out earlier, that's thrown-away time. They don't want you to wait in line. The reason that you do sometimes wait in line is that there are surges of demand that they can't anticipate. The ones that they can, such as people coming in between, right after work, it's very expensive then to hire an extra person for just a short period of time. So, they can't easily solve that problem. But, in general, stores don't have long lines all the time. That just doesn't make sense. It means they should raise the prices of all the goods a little bit, even, if that's the case.


Russ Roberts: But I want to go back to the point you made earlier. Because, I made a long speech, about 10 minutes ago. And, you responded by changing the subject. And that's because you are very polite. Which is very impressive. But, I actually wanted you to push back on that. And since you didn't, I'm going to push back on it with what I learned from you earlier in the podcast. Because, I wanted to--I wanted to reconcile the way you were thinking about this congestion problem with the way I think about it. So, in my model, my--the way I'm thinking about it--it's like everyone's got the same--let's just start with this can-opener assumption. If you don't know what I mean by that, just google economist jokes and can opener and desert island. So, I'm going to make this economist's simplifying assumption: everyone kind of has the same value of time, to show that when I put the congestion tax on, everybody is worse off. People who don't drive; and the people who do drive. They've all been punished a little bit. And the gains are being captured by whatever the government spends the money on--whether that's food stamps or Social Security or improving the sewer system or whatever it is. And since the tax falls heavily on one small group--drivers or would-be drivers--that seems unfair. Immoral. Not so attractive. Even though it adds to the net benefit to society. And that's why I'm not a utilitarian. One of the reasons I'm not a utilitarian. I do not like that tradeoff. I don't think that's what government should be in the business of doing. But, your point was, the way I understood your point was: Well, but people have different values of time. And what's actually happening when you put a tax on, is you are persuading some drivers to give up and stay home. And normally in a market--and this, I think is very deep, and I, this is something I didn't appreciate--in a market, the people who want something badly, compensate the people who don't want it as badly. That's your widget story. You make the widget--you are willing to accept it costs you 50 cents [$0.50] to make it; you are willing to accept, given the value of your time to sell it, you are willing to accept, I think you said, say, $1.50, or a dollar.

Michael Munger: I think a dollar.

Russ Roberts: A dollar. And I love widgets. So, I'm willing to pay $5. So, $2.50, $3.00--any price in between a buck fifty and five, you and I are both going to be better off. And I'm compensating you, effectively, for giving up the widget. What happens in the road case, because it's not a market--and this is what, the humor of this is that you claim you learned this from me. I didn't understand your point. And this is what I think--this is what I'm learning what you learned about from what I said that you learned from me that I learned. Which is: That, when you put the tax on, nobody's being compensated. Not the people who stay home. Not the people who drive, either. So, I'm the--let's say I'm the high value of travel. Meaning, it's really important for me to take the trip. And I have a high value of time, so even though there's a congestion fee, I get on the road. I don't compensate anyone who got, who chose not to go on the road. I think that's your point. And that's very deep. I'm also adding the point--the reason I was reacting so strongly in my monologue a few minutes ago was that, 'Yeah, but I'm not getting compensated, either.' I'm getting punished, too. But there will be people who are not punished, who are better off. Who have high value of travel, and very high value of time. And they have--the tax induces other people to get off the road so that they can enjoy that road with their high value of time and not be punished by a lot of congestion. And what you are saying is that, 'That's not the way it usually works with an exchange.' Is that right?

Michael Munger: Absolutely. And it--the fact that we use, that the people who advocate for this use a sort of exchange-based language--we want to make it efficient. And as it is, what we want to do is deter the people who have a low opportunity cost of time so that the people with the high opportunity cost of time can use money to buy their way out of the deadweight loss of having to wait in line--in traffic. And, it's not an exchange. It's just as you said: The people who have, because of the high price of the congestion tax, if they don't drive, they are staying home. And that should be a deadweight loss that we account for also. They are not being compensated. In the exchange metaphor that we started out with the widget, and again, this is what you just said, that means that, 'Yeah, I don't have the widget any more, and that makes me sad.' But I'm only sad a dollar [$1.00]. But, I'm happy, $2.50. So, I'd like to have the $2.50 and the widget. But, I'm pretty happy to give you the widget and have the $2.50 in payment. But, that's not what's happening with the congestion tax. It's just a net loss to me. I'm not being compensated. And I think that, not taking trips, not going to the park, the things that people who, because of the tax, now it's not worth it for them to undertake the trip, should at least be counted. Now, usually we count that in terms of ability to pay. And so, I guess I'm making sort of a point that many of our friends on the Left would often make. What we're really doing is not valuing--we're not rationing by value of time. What we're doing is rationing by the ability to pay. And, so, rich people are going to be the only ones who use these roads, at a time when there's a congestion tax. And that's the objection. That, if you look at the cities where the use-congestion tax--so, Singapore, London, Milan, Stockholm--that's the objection that everyone makes, is 'Well, only wealthy people are using this road.' That might be okay if they were compensating the people that were being kept off. But they are not.

Russ Roberts: Well, my point, which is, a variant of this, is that: There will be some poor people on the road, who really want to use the road at that time--because they have to get to work--

Michael Munger: Yeah. That's right. And they benefit from that, actually.

Russ Roberts: Not necessarily. Well, they don't. My point is that they might not. Because they have to pay the tax. And that offsets some of the gain that taking the trip had. And, it's true they get there quicker. That's nice. But the tax is very large, to get enough people off the road so that it's not congested. It's not obvious that they are going to be better off.

Michael Munger: Well, not compared to the old system where they would have been on the road anyway. Not pay the tax and have to wait longer--

Russ Roberts: A little longer. But they have a low value of time.

Michael Munger: But that's no longer available to them. That's taken away.

Russ Roberts: Correct.


Russ Roberts: I want to go back to the widget example, because I think that really helps me see your point. Which is that, what a congestion tax is like, is the government comes--and it is a little more complicated than this, but it's close. What a congestion tax is like, is the government comes along and says to they guy with the widget, 'You know, I don't think you value it that highly. I'm going to give it to this other guy.' Hmmh-heh. And then, the guy who made the widget doesn't get any, doesn't get compensated. The guy who gets the widget gets to enjoy it. But, actually, we're also going to make the guy with the--who gets the widget pay a tax also. Hmmh-heh.

Michael Munger: Yeah.

Russ Roberts: We'll pay a tax. We're going to take it--We're going to--sorry. He's going to--it's more like this. He's going to pay the $5. In fact, he's going to pay a little more than $5. Not a little more. Excuse me. He would have paid $2.50 in the private equilibrium in the market. Now, the government charges him $4. He's still better off; he's happy to get the widget. But the government takes the $4; doesn't give it to the guy who made it. And, um, the guy who made it, and the guy who gets to enjoy it are both worse off than they were before.

Michael Munger: Yeah.

Russ Roberts: The difference is that there's no supplier, in the case of the road, except the government, in some sense. There's no residual--what we call in economics, the residual claimant. In the case of the widget, we all understand that if widget owners, widget makers, don't get paid for making widgets, they are going to do something else. That's not going to work for very long. So, no one thinks of that as an attractive option, at least literally. 'Oh, well, let's just take the, we'll just confiscate the, expropriate the stuff.' Most people realize after a while that doesn't work so well. Some don't realize it for a long time. But, in general, the market system maintains that incentive for the supplier to make the stuff. The road is a little trickier, because there is nobody profiting from the road directly. There is nobody who owns the road to collect the toll. The government does. And so, to the extent that the government, whatever you want to call the government's decision-making process--we know, from our previous conversation about choosing in groups, in your book, that that's a complicated phenomenon. It's not simple to say, 'Oh, the government will just make this decision.' Or 'that decision.' Depends on what the institutions are. Depends how the political power of people who are, who have a high value of time, political power, the people who get the money spent on them ultimately. It's going to be messy. Right? But it's clearly not like a supplier who gets a profit and has an incentive to make a great product.

Michael Munger: Yep. And I--I had not realized that I should, but I will from now on, object to the sort of metaphor that was used as if it was an exchange. Because it was the--in preparing for my notes, it's eerie, actually. In preparing for my notes for this, I actually had really had written out the second part of the example of the widget, a social planner shows up--

Russ Roberts: Hmmh-heh--

Michael Munger: and says, 'All right. I think you value it one. And you value it five.' And let's suppose she's really good at it. She actually gets it right.

Russ Roberts: Yeah.

Michael Munger: The police come to your house and take the widget from the person who values it one. They give it to the guy who values it five. But they charge him for the service. That's not an exchange.

Russ Roberts: Yeah.

Michael Munger: They are actually both worse off than they would have been if they would have made an exchange. Now, you can say--and I think this is right--a difficulty is the transactions costs are so high, there's no way that they could have found each other and made the exchange. I cannot find an individual or the 5 individuals I need to pay to stay off the roads so I can go quickly. So, the exchange is impossible. But the analogy to an exchange--

Russ Roberts: Yeah--

Michael Munger: for people who advocate for congestion--that's wrong.

Russ Roberts: Yeah--

Michael Munger: This is something else.

Russ Roberts: Yeah, that's great.


Russ Roberts: So, that leaves us, or me, any way, you can speak for yourself, that leaves me with the challenge, okay, so: You're so smart. You're so high-falutin' about your principles with respect to a congestion tax. What do we do about the fact that there's traffic in the city? Because when people say, 'You need to widen the roads,' I always say, when you do that, you encourage people to live closer to the city or more people to move there. That might be good. That's a very complicated calculus. You could try to make it right. But it's not obvious that that is an improvement. It could be. But it's not obvious that it is. And the improvements in terms of time savings tend to be short-lived. Similarly with public transportation. 'Well, let's build more public transportation.' Well, that has a short run effect. But eventually it encourages people to live where the stations are; and that then crowds out other roads around there. And if it frees up people from the main roads, more people move in to be able to use the main roads. And you could say, 'Well, that's good,' because there are more people in the city enjoying the benefits of the city. And that's true. Which is good. Which is why sometimes you should build roads. Which is why you should build public transportation. But it's a cost/benefit analysis. But, the standard economist's answer of, 'Well, we should need to tax the roads,' to me is unacceptable and false. It's not morally obvious that taxing roads is a good thing. But it does then leave us, put us into no-man's land without an easy solution. What are your thoughts on that?

Michael Munger: I have a friend, Jason Scheppers, who works on congestion pricing for the state of Texas and a couple of universities in Texas. And I asked him just that question, Because I was confused, myself, that, 'Allright, you're so smart. What should we do?' And, his claim is that Newark is the city in the United States that is closest to having a congestion tax make sense. At some point, if you have enough congestion, you might be that you want to use this blunt instrument. But, in fact, most people pay the full average cost of being on the road. And the general equilibrium effects, which is what the fancy way of saying what you said, that people will take into account of the costs of having to commute. And they'll make location decisions that are based on that. Well, since that's true, the congestion tax is not an improvement. We're probably better off saying that we have a plan for the way that the--and by plan, I mean we have an announcement about this is that roads are going to be for the future so people can predict. They make location decisions. And yes, they pay, then, the amount of waiting costs that are involved in traffic. Congestion--there is no evidence that congestion pricing is an improvement in the way that it's often advocated for. And I have to admit, I was surprised at that conclusion. I've, I always accepted the economist's idea that it would be better if we priced for it. Now, let's--there is one caveat to that. If it were possible to have private roads--and I think this is an interesting question: it's too big to raise at the very end of our podcast. But: Why are roads public goods in the first place? Public goods are, have the two properties that they are rival in consumption, nonrival in consumption. And nonexcludable in provision. That is, it's difficult, it's expensive for me to charge you to use it. And, if you use some of it, there's no reduction in the amount available for someone else. Neither one of those things is true for roads any more. So, my private road, sort of unexpectedly be an alternative. So, instead of public roads with congestion pricing, private roads. And like you said: We'll just build another--on top of 5th Avenue, we'll build another road. Heh, that's 50 feet up. And well, okay; we're making stuff up. But private roads might be a solution. And a number of places have done that. Toronto, sold off one of its central arteries to a private company. Maybe it hasn't worked out very well. But the dynamic pricing mechanisms where you were able to charge something where you were covering the average rather than trying to focus on getting the right margin, might be an answer. So I think that's not a very satisfactory answer. That's the best thing that I could come up with. I've changed my mind about congestion taxes. I think they are a bad idea.


Russ Roberts: Yeah. I don't want to go too far on that point. I would say they are not a good idea based on the standard arguments that people usually make--"they are efficient" or there's a negative exernality. Again, the real--I think the reason the reason the intuition breaks down, for the economists in the audience, is that usually a negative externality is me imposing costs of everyone else, and not taking into acount. The driving is, everybody is imposing costs on each other. And it's not the same--it just doesn't go through the way it normally would. When I tell economists that congestion tax makes drivers worse off, they always laugh: 'Oh, you didn't do it right.' Because we always are in a case where I impose an externality on someone. Pollution, say. Well, then the people who were breathing the air are better off. Yeah, they are. It's not the same kind of--it doesn't work that way. It's not same. It's not analogous. I think that's one point to make. But coming back to your point about private roads: I think you are ultimately getting into the same challenge there because of the existence of transaction costs. And I'll say what I mean by that. If we took the Beltway--which is the 495 that goes from the City of Washington, which is horribly congested numerous times during the day--sometimes on a Sunday afternoon at 3 o'clock, even, but certainly with great regularity at rush hour on weekdays. And we said, 'We're going to sell this to a private person. A private entity. And we're going to let them put a toll on it.' Well, they would put a toll on it for sure, because it's cost effective, the technology. So, to get on the road you'd have to have the box or the EZPay pass, whatever it was. And then they'd charge. And they'd be able to extract money from all of us who had made our decisions already about where we live. And to get around. And they would make a lot of money off of it. And they would presumably--they would set the fee--I think this is true, we'd have to think about this more but I think it's a result, standard result--they'd set the fee so it wouldn't be [?]. So, at times--they would charge, all the time, but they would particularly charge a lot when it was high demand. Say, rush hour times. Morning and evening, when people are going and coming back from work. Well, that's going to end up being efficient. Hmh, heh. Because, if they sold that to the highest bidder, people would pay a lot for the privilege to extract money from drivers. But we all understand that wouldn't be a nice thing. It wouldn't be a good thing. It wouldn't be attactive. Even if it were true that it got rid of traffic. Because of the opportunity to [?] price. And if you say, it's true, that drivers are worse off; but look at all the money that the government got from the auction. They can use that money to compensate the people who are worse off. Well, they could, interesting theory. But most of them won't be able to do that in a way that will actually work. And if you start thinking--well, they'll--I don't even want to explain why it's hard to do. But it's hard to do. So, it seems to me that the fundamental challenge is that getting to place to place--because you can't build 50 roads on top of each other at the same--tje second road isn't as easy to build as the first. It has to be above the first one. It has to have higher cost. And it's going to be ugly. So, there's no attactive--my conclusion to this is, I think, you just have to live with the fact that when a lot of people want to live near each other, there's just going to be congestion. And there's no way that, as smart as we are as economists--we're somewhat smart--we can't "solve that problem" through pricing or private ownership. The standard ways we solve problems. It's just the fact that there's a bunch of people trying to live near each other and they pile up. And there are ways to make it a little bit better. I'm open to the possibility that HOV lanes maybe are an improvement. Or, maybe it's possible that a congestion tax could be a good thing. But, you and I would be the first to point out that there's no reason to think that the government is going to set the prices arguably to efficiency. It is just a tax. And once you say it's a tax, the level that the tax could be set by the political process, not by the engineers who have somehow solved that credible set of informational challenges you mentioned earlier.

Michael Munger: I'm afraid there's no way around that. And so, I raise private roads partly tongue in cheek. Because, one of the tests that I always propose to the people, is they have some idea that' 'This is the way that markets are failing. And I could do it better.' 'All right. Well, why don't you do it? You could start one yourself. There's a tremendous profit opportunity.' As far as I know, there are no examples of major cities where a private company has come in and said, 'You know, we are going to pay all the costs; we are going to borrow the money.' I'd be interested to know if there were any of these. But, 'We're going to borrow the money; we're going to build a road'--and we're going to, and it's an alternative to the existing road structure. And then, 'We're going to charge for it.' So, I don't think that private roads in the usual sense are actually viable. So, while I changed my mind about congestion tax, it might still be the best solution that we have from a bunch of imperfect ones. Recognizing all of the difficulties that we've brought up. But, you might very well persuade me that a congestion tax that works in a way that we haven't yet specified might be the best thing that we could do.

More EconTalk Episodes