No, ads won't pay for your robotaxi ride -- but your employer might, and that has big consequences


Most of the press reported a research report from UBS securities claiming Waymo is now worth $75B to Google because it is poised to dominate the robotaxi business. In addition to this, it claimed that business would be $1.2 trillion by 2030, with an additional $472 billion for "in car monetization." (Total Google revenue was $110 billion in 2017.)

(I can't find the UBS report itself, if you can send it to me, please do.)

In general, I agree with the first part, and have been saying so since the beginning. For investors, the main problem has been that this mega-huge business that anybody would like to invest in has been buried inside the $750B market cap of Alphabet/Google itself.

But I want talk about this idea that in-car monetization will be 40% of the revenue (or 28% of total revenue) of a robotaxi business. This reminds me of the very frequent claim that "Rides will be free, paid for by ads." In fact, when Google first came out of stealth with the project, this was a very frequent suggestion. After all, Google is pretty much 100% an advertising company, so if they're making a car, it must really be about advertising. The UBS forecast pretty much puts that down -- transportation is a much bigger business than advertising. Your robotaxi will be more like an extension of your living room or office. Apartments don't get 40% of revenue from "in-room monetization."

But will ad revenue, or other "monetization of the data" from transportation become an important factor, even a dominant one? Could it be the most profitable aspect of running a robotaxi service as some suggest? Or does it make sense that rides will commonly be free, supported by some sort of advertising?


Trends can change, but for now the answer is it will not. The transportation industry is a much larger industry than the advertising industry. That doesn't mean that can't flip, but most proposals involve the tail wagging the dog.

Much more likely are rides subsidized by destinations, and most of all by employers. If most commutes are paid for by companies, that has big consequences.

Driving costs $20/hour today

Today, most sources suggest that operating a private gasoline car costs about 50 cents/mile for a medium car (plus parking.) Older cars can be cheaper than that, and fancy new cars are more expensive. We like to price driving by the mile, but it might make sense to price it by the hour instead. (The age of most motorized vehicles, like planes and boats, is measured in engine hours.)

On a basic level, with the average speed of 40mph for cars in the USA, you can multiply that out to $20 as an average cost per hour. In fact, it's pretty clear that fast highway miles don't cost nearly as much as slow city miles. Gasoline cars are more efficient at highway speed, and on the highway they run in one gear, at near constant revs, without much braking and acceleration, well cooled. The city involves starting and stopping which don't just make gas mileage worse, they add more wear. Some types of wear (like wear on the interior) are very much by the hour.

It doesn't matter what you exact calculation is, $20/hour is a pretty decent proxy for the average. Electric cars are cheaper. They are more efficient at low speeds than fast ones, use cheaper energy and have lower maintenance. The main thing that wears out is the battery, which wears out based on how hot it gets from things like high acceleration, regen braking and fast charging, as well as the number of discharge cycles.

I've predicted that robotaxi operation will be much cheaper -- we'll get into that below.

Advertising makes around 60 cents/hour!

Let's start with TV advertising, which gets the largest share of ad dollars at 38%. Running a 30 second ad spot varies in cost, but $20 per thousand impressions (CPM) is a typical number. That means the advertiser pays 2 cents for every person watching the program. (Not all will be watching the ads, as some leave the room.)

If you watch an hour of TV it will have about 30 of those ads spots, taking 15 minutes. So the Network earns 60 cents for bringing your to their sponsors. 60 cents an hour. By the time it gets from the network to the studio and the actual creators, it's a lot less. You probably get well under 40 cents of programming in exchange for watching 15 minutes of ads. (You can't buy the programming for 40 cents from the studio though, advertisers get a wholesale price and a viewer premium by making it free.) Well under minimum wage.

Some types of advertising do better. Print media actually get three times as much revenue per hour of your time. That's because it's so quick and easy to flip past the ads in magazines and newspapers as you read. If you like an ad, you pause on it and spend more time with it. The rest of the media tend to have similar ratios of time to revenue.

There are specialty types of promotion which do better. If you are advertising to billionaires, you're going to pay a lot more because you earn much more, but the billionaires are cautious with their time. The internet is a strange beast too. It has the same ratio for all internet use, but the distribution of money is not even. Your money goes to companies like Google and Facebook, but you don't spend that much of your time on Google. You search quickly, glance at ads and click on them enough to make Google very profitable, and then go on your way to free sites.

The other most lucrative types of promotion are the ones that trick people -- such as the timeshare pitch session.

Something bringing in 60 cents/hour can't pay for a $20/hour car ride.

Ad-driven ride

You don't, and won't, spend your whole time in the car consuming ad-supported services. And when you do, it's not very likely you will consume services and media from the ride vendor. Most of the time you are going to be just using your phone, consuming media on it, or doing all the other things you like to do on it. That's what passengers sitting in cars and transit do today.

The robotaxi will have a big screen and nice speakers in it, but I predict these will mostly be peripherals to your phone. When you want to watch a movie, you will be as likely to use a subscription service like Netflix as to watch ad supported video. Whatever you want to do, you will prefer to get it through your phone, where you have everything set up the way you like it, and have a user experience you know and selected.

This is even true for travel related services. When you are riding and feel like getting a bite to eat, will you go to the system of the taxi you're in, or will you just use your phone? Your phone knows where you are and unless the car deliberately locks it out of this information, it will know as much as the car about where you're going. I don't want to ride in a car that refuses to tell my phone such things, but since I almost surely ordered the ride using my phone, it knows anyway.

I can't imagine people preferring to trust the "free" advertising-biased recommendations on where to eat coming from a car's console, over the restaurant recommendation tool they already like to use in their phone, which is also free. (The phone apps often have their own ad support.)

So just what are these services you will use in the car? Of course, my imagination and yours are limited, so I am sure people will dream some up. Perhaps some games. But it's hard to see how the phone people won't also think these up, or how the car-based services, which change function and UI every time you use a different type of car, will get that much of the pie.

Of course, with 2 people in a car, you could get twice the ad revenue. But even though the roof of a bus or subway is lined with ads for the dozens of people on board, the advertising revenue doesn't dent the cost.

Specialty hardware based services

If a car contains special physical features, people will be keen to use them. I'm sure there will be cars designed just for having a party with friends. Or sleeping in. Or picking up kids at school. Or even playing an interesting new game. People will pay for that, and maybe in certain cases they would let the special feature be paid for by advertising. We'll see.

Free ride to the sponsor's establishment

Much more likely will be sponsored rides with a specific goal. A nice restaurant might offer free taxi service to and from the restaurant. It's not going to be expensive to offer this. This might be particularly appealing for bars -- eliminating drunk driving risk, as it could. It's possible that offering such a free ride might become an industry standard -- if all the other bars offer free rides, you might have to as well. It will be a good way for ride companies to lock in business by making exclusive deals with particular locations.

Of course the rides will not be free, you will pay for them in the drinks, and people will know that, but people have a bad habit of easily being tricked by such things and even getting used to them. I mean many places have free or validated parking, even though parking can be quite an expensive thing to provide, but customers come to expect it. Such rides might try to "lock you in" to a particular bar -- leave for another bar and you don't get your free ride home -- but then the other bar probably offers that ride to get you over. Restaurants don't expect you to hop to a rival restaurant so they don't have as much problem, and you want the freedom to wander afterwards anyway.

Stores might have a harder sell. Most people like to visit more than one store on a shopping trip. Malls could make a comeback in this way. With bars and restaurants, you know people are going to eat and drink, and can even insist on it: "Free ride if you order food." With a store, a person might take the free ride, browse for 5 minutes and head to their real destination. Stores can't easily force you to buy.

I have suggested that the neighbourhood elevator -- a local shuttle to bring neighbours to the shopping street -- might well be free and sponsored by the merchant association. That's an easier sell than a ride that binds you to a single store.

Free ride to work

Employers might offer free commute rides. First of all, it's a nice employee perk, and many employers already offer free bus service to the office now. (European companies always give management employees a free leased car, too.)

Employers can even try to make it efficient, when they notice two or more staff that they could put in a carpool, they would do it. Everybody's going to the same destination, and it's a free ride from your employer, so you will not complain if the car pauses to pick up a co-worker. (Unlike today's carpools, it will never take you out of your way, since other robotaxis can bring the co-worker to the meeting point along your path. You might have to leave the highway.) Employees can even push the "No, I'm running tight for a meeting, don't carpool me today" button.

These are free rides, but not ad sponsorship. The employer is probably happy you will do work in the vehicle, of course. This is just business, and since if you paid for your own ride it would be with money you got from your employer, it's a wash, and actually a big win for tax reasons if the taxman doesn't try to stop that. (Further notes say that private taxi rides would be taxable benefits but van rides might not be taxable.)

Employers -- and other locations -- have a huge incentive to offer free rides: They no longer need parking. This can be a major cost saving in denser areas, and growing companies surrounded by parking lots suddenly can have space to build new offices.

Consequences of free rides

Free rides to destinations do not count, in my book, as "in car monetization." It's just another way of selling taxi service -- you sell it in bulk to employers and establishments.

As I described above, if it becomes popular for a class of destination to offer free rides, that will push all the others to do so. Only the more educated consumers will understand deep down that the ride is not really "free."

If this does become popular, it could make a big change in the economics of car ownership. There is much debate about how many people will want to own their robocar compared to those who live the all-taxi life. But if half your rides -- especially your commute -- become expected to be free, it pushes people strongly away from ownership. It also pushes them away from even ownership of a manual car. The free commute would immediately drop ownership in most households from 2 cars to 1. It might overcome the fear of the vehicles.

We could see negative consequences too. Companies might make this Orwellian -- it allows them more control of employees, and records when they come and go, can even force them to travel at certain hours. (Want to come in late? Pay for your own ride.) You will have far fewer rights on a ride you didn't pay for, and you'll put up with more problems. Just as having employers pay for health care has come with lots of gotchas, we might see some parallels in transporation.

Governments could see powerful levers if companies are paying for most of the rides at rush hour. Levers they could never apply to private drivers can be applied to companies. This could mean better traffic management and lower emissions, but also to over-control of the city.

But the ride will get cheaper

As indicated, I forecast that riding in a robotaxi will get much cheaper than $20/hour. Perhaps as low as $4/hour in the city, and $10/hour on the highway.

I have a detailed article on Robotaxi Economics exploring some of these costs. Be warned though, Uber charges 40 to 50 cents/mile right now (20-25% of your fare) and doesn't make money, and Uber pays no car or driver expenses -- only insurance, operations, software, marketing etc. That suggests the problem is hard.

But it's still not enough. We need both transportation to get 5x cheaper, and we need advertising to get 7x more effective, before an urban ride can be paid for by ads, and you'll have to spend the entire ride using the service to get that free ride. Just to save a buck or two.

And you also have to somehow pay for what the ads are embedded in. Unless you imagine people would just sit staring at ads for no other purpose than to get ride credits. I hope not.

But isn't data the new gold?

Advertising can't pay for rides, but some people write that having all the data that comes from transportation will be a true source of value. It will have some value, but it's not really a lot more data than is already learned by companies like Waze, Google and Apple Maps, cell phone companies and companies like Uber and Lyft. Nice, and a good niche business, but again it's the tail wagging the dog.

Can advertising get more effective?

Many people have hope that as ad targeting gets better, advertising can become more effective, and get higher revenue per impression. The theory is that if you only see ads about things you are truly interested in, you are much more likely to buy the products, and the advertisers will pay more.

That works to a degree. It's one reason that web search ads work -- you are actively looking for something, and you get ads for the thing you are looking for.

That's different from general advertising. Here, people imagine that you might watch a TV program, but because the system knows you are in the market for a car, and trying to plan where to vacation this summer, and have a small baby, you see ads only directed at these known interests. This can be creepy, or it can be designed to be delivered only by your own computer that gets to know these things about you but not report them back to HQ.

I have doubts this can work. For one, I find ads that are related to my interests to be more distracting and thus more annoying. Only rarely do they inspire thoughts of, "Oh that's handy, I am looking for a car and there's an ad for Ford."

Most of all, though, I don't think you can boost inequality that much. Ads exist to manipulate our buying choices. The better they are, the more sponsors will pay, and the more they manipulate me. Every dollar sponsors pay is often a dollar of negative value to me. We may not realize it consciously, and as such advertisers do get away with it. But there's a limit. To be effective, an ad has to make me buy something I would not otherwise have bought, or to keep me buying what I already buy and not wandering off to the competition. If it's not doing that, why would advertisers spend money on it?

As such, I believe there is a cap on the value of an ad impression to an ordinary middle class person. The higher the price for the ad becomes, the more it's manipulating the target. People should resist that. (This is different from the higher value of ads delivered to wealthy people. Those are expensive because those people spend more, not because you are better at manipulating them.)

Consider the extreme -- advertising so effective that people always clicked on it or always bought the products advertised. Sponsors would pay a lot for such ads, but targets would start to resist them for just that reason.

There are some counter examples to this. I will examine this more in an article that is more about advertising than about robocars.

Are there any forms of advertising that could be lucrative enough to pay for a ride? Let us know in the comments.


The obvious point can be imagined by looking at what industries actually do the heavy advertising. My guess is that cars and mobile phones are in the top 5. Do people really imagine the robotaxi service existing to support ads trying to convince people to not use robotaxis? Or that robotaxi fleet operation is sustainabile by convincing people that the other robotaxi fleets are inferior? Seems absurd.

I don't think pharma advertising will pick up the slack since a world without human drivers will be a world that sells a lot fewer anti-anxiety pills. :-)

Well, we spend a lot of our money on cars and phones of course. Though other big advertisers are beer and cheap things. The car ad might sell a very expensive car, but only a few percent of the viewers are in the market for a Ford. On the other hand, a large fraction of the audience might buy beer. Beer ads don't cause people to change brands too much but they make them feel good about loyalty to their brand, and thus less likely to change.

But I just don't know where you are going to see these ads. When I am in my robotaxi, I will either be looking at the world going by, talking to friends, or staring at my phone, or working at the bigger computer, or watching a video that comes from my phone on the screen, and there will be no ads in that video. If they try to interrupt any of these things with an ad I am using a different service next time.

Just discovered your blog through Twitter, and I think you have brilliant points here. I will add that I've run my own calcs, and I do think that the mix of no-frills two-seaters in dense urban areas will be a lot higher than most expect, in part because these cars really can be ad-supported: Use shoots up dramatically for 'free'. I am trying to wrap my head over what would win - the promise of a peaceful ride with no choice in destination, or ads + choice, and I frankly can't see the end of this tunnel. I would personally take the peace, but I am not near the center of the bell curve. And the cost of complete freedom (paying the fee myself) is frankly trivial for me.

Or perhaps all of the cars will be paid for by the state and feature ads for Dear Leader. To be perfectly honest, I am not sure that capitalism or democracy are going to make it to 2025 at this point. All we can do is spin the wheel, and see what numbers come up.

While that no frills 2 seater is going to be very common and very cheap, it's not cheap enough to pay for with ads.

Can you paint the scenario where i get into this car, presuming it costs about 25 cents/mile to run (which is not happening for some time) and tell me how it becomes free with ads? Note that I don't view destination supported as ad supported, it's two different things. So if you mean destination supported, so you get in because it promises to take you free to a sponsoring business, I can see that happening. I don't see it taking you anywhere within a general zone because you will watch ads while you ride. It just doesn't pay out economically. I can see it taking you anywhere in a "business district" if that district has a retailer's association which pays for the vehicle. Would their be enforcement, ie you must get validated, like parking, to get the ride for free? Or will they let you ride free even if you don't go to the store? (Say you live or work near the store and realize you can always score a ride home.) If people are identified you could limit them to one free ride a month or something.

So how does it play out? And remember, when it's this cheap, a paid, works-for-you robotaxi will also only cost under $1. What will you be asking people to put up with to save $1? Not that they save it, the products at the store will cost more than $1 more because of the free ride, of course.

"UBS securities claiming Waymo is now worth $75B to Google"
In comparison, a May 15, 2018 report by Morgan Stanley values "Tesla Mobility" (what Tesla refers to as "Tesla Network") at $95/share, which equates to about $16B, with about 179M shares outstanding. Morgan Stanley is delaying its forecast of the launch of this service by one year, to mid 2019.

Morgan Stanley values Tesla's auto business at $196/share, Mobility at $95/share, and energy and solar at zero, for a total of $291/share.

"Just as having employers pay for health care has come with lots of gotchas"

I'm not sure what you are referring to, but I doubt that this is true in general. Certainly in my experience, health care is a catastrophe (meaning people die who shouldn't) where it is private and this is simply not an issue if employers and/or the state pays for it. Nor is the quality worse (if anything, probably better, and less malpractice because a surgeon can't perform an unnecessary operation to enable him to buy a new car).

Thanks Brad, interesting idea regarding employers paying for getting their staff to work. Many of the restrictions that could be placed on this service are already reflected in employer provided cars with GPS trackers. Many years ago my company vehicle came with no strings attached, perhaps because private use was so difficult to measure it was not worth the company trying. GPS technology changed this and soon the accountants were pointing out cost savings by placing more and more restricting on private use - at least this was my experience.

Another cost that has been too hard to measure is individual contributions to traffic congestion. As a hypothetical, say an employer offers free travel to work using robo cars. If several people applied for a job but the slightly better candidate lived 3 times further than the next best candidate, it costs the employer almost nothing to go with the slightly better candidate. The employer pays nothing for the employees travel time nor the extra contribution to congestion, the true cost of which is spread over the general population. In other words the cost is externalized so their is little incentive to deal with the problem at the source.
It is possible that GPS technology and sophisticated road pricing could transfer this cost back from the general population to those making the actual decisions that contribute to the problem. Putting a true cost on something is an effective way of reducing waste, in this case millions of hours lost due to traffic.

I can't say I have ever heard of an employer taking an inferior candidate because they live closer. Maybe for a low tier job -- but actually in the Bay Area many lower paying workers can only afford to live in the Central Valley, with a >1 hour commute, while higher paid workers can live here. But many companies here now provide transit for workers from a bit over 1 hour away, in fact hoping to be more acceptable to that distant worker.

I have outlined other problems for congestion. But in fact, employer robocar I think will be very good for congestion because the employers will create on demand car pools. While on demand car pooling is always good for congestion and reducing cost, there is a bonus when it is people from the same company, because they can talk to one another about work while on the ride, which you can't do on a city bus or UberPool.

Companies might even be the first to implement proper robocar transit. A company in Silicon Valley like Google would, rather than have bus routes into the city, bring people from all over San Francisco to a common point where a bus (robotic or otherwise) is waiting to take them nonstop to the office. In such a system, employees go door to door, and they leave when they want to leave, it's 95% as good as going in a private car, but vastly more efficient. And social.

Under current law in the USA, providing your employee with a taxi ride to work would be a taxable fringe benefit, and would have to be included in the employee's W2 (and would be subject to social security and medicare taxes). For this reason alone, I don't expect it to be common. Also because there's really no benefit to the employer by doing this.

Oh yeah, parking. Well, employers in places where parking is an issue often don't provide free parking to their employees anyway. So it's up to the employee to pay for parking or to take public transportation anyway. Again this is the US. Maybe it's different elsewhere.

But basically, unless there's a tax advantage, and I doubt there will be so long as the Republicans are in control, I don't see the point. Employees can pay for robotaxis themselves. They can even join robotaxis pools with other employees themselves. And then each person can pay what's appropriate based on where they live, instead of effectively having the people who live closer subsidize the people who live further away. (And when you think of it that way, giving a tax break to employers that subsidize increased transportation costs probably won't be something the Democrats will rally around either.)

Clearly people can pay for their own transport. At most employers that is what they do. However, it it became the case that an employer could offer a robotaxi pool ride to employees at a cost of just a few dollars per employee per day, I think many of them would offer it. They would offer it as a useful tax-free perk, and a way to eliminate the need for their own parking lot.

It wouldn't be tax-free in the USA. It'd be a fringe benefit that doesn't fall under any of the exceptions for tax-free fringe benefits. I also don't see how it could be only a few dollars per employee per day. You get some benefit from essentially forcing all employees to use it, but not *that* much.

Maybe I am mistaken, but I don't think that Google's bus service, which provides a free ride to the office, and the others like it, are taxable benefits. There has been talk of changing those rules. If a solo service would not be permitted, it might be that employers would try to always carpool, and that is something large employers can easily do but smaller ones would have trouble doing, without pooling with outsiders.

Under the rule that Google is probably using, the vehicle "must seat at least six passengers, with at least half the seats filled by employees, and be used at least 80% of the time for employee commuting."

Companies which already choose to take advantage of this tax perk will likely move to driverless vehicles to offer it. And at some point the cost of driverless vehicles will drop to the point where some companies on the margins might be able to offer it. Unless the tax break gets dropped by then, which seems likely.

Thanks for looking up the rule. Of course, they might get the rules to change -- employers are powerful and if they can pitch the advantages for congestion and energy it could be a reasonable sale. Otherwise it might be a taxable benefit, which is less exciting.

However, they could offer a combination of a very short ride to join up with a 6 person van which takes the employees to and from the office. In a manner similar to my plan for future transit, the employees would be picked up in small robotaxis and taken a short distance to meet up with 5 other employees who happen to be leaving around that time. The short ride might be a taxable benefit, but the van ride would be non-taxable. (perversely, it could not be used in the rest of the day, a rule that could probably get changed.) The key difference is today, employees ride buses from fixed bus stops, and must meet one on a schedule. In the system I describe, the employee just declares on their phone that they are ready to leave (or will be at a specific time) and the system finds 6 employees leaving at roughly the same time in roughly the same area and makes it happen. If it can't, the employee travels further to find a bus or van, or goes solo all the way, but if that doesn't happen so often it works out.

Even if it's a taxable benefit it's a good one. The employer probably gets a very good price and the employee would need to pay for their own ride directly without this, with post-tax dollars.

As I said in my initial post, I don't see the rule changing. Most Republicans would rather get rid of these sort of "green" tax expenditures rather than expand them, and most Democrats are not going to be in favor of expanding the break in a way that reduces the incentive to carpool.

As for picking up the employees in small robotaxis and taking them a short distance to meet up with 5 other employees, I can definitely see employers giving employees vouchers that they can use for mass transit, and that mass transit eventually being driverless. (And maybe it'll wind up taking just those 5 employees, or maybe it'll wind up taking 5 employees from one company and 5 employees from a nearby company. Either way, it won't need to sit around the rest of the day, though there will of course be times of high demand (rush hour) and times of low demand where most of the vehicles are parked.) The vouchers won't be good for the short solo trips (there are too many administrative costs to do it if it isn't tax free), and maybe the vouchers will only cover the cost of the average distance employee rather than all employees (the ones living in the boonies will have to supplement the vouchers with their own money). But I can see it being offered to some extent.

So yeah, I think you're largely right, and I think the specific details of how the employer will pay for it are going to be a bit different from how you explained it.

Yes, some will use vouchers. However, they find an attraction to having "company only" vans. First, they can design and configure how they work. Google's buses are nicer than a typical commuter bus, and have work tables and Google wifi access (though not secure, but all Google devices use Google VPN anyway.) Employees can feel free to have work conversations, which they can't have in a vanpool with other people. They want their staff to feel special. And people do love the buses. Of course, what is different is today, the public transit is much worse than the company bus. In the future, the UberPool service might be quite similar in quality to a company van, and more competitive.

I love the way you use basic economics to ground the arguments in reality. I would suggest that cultural factors should be considered also. I used to work for a tech company headquartered in Great Britian, and was astounded that first level managers had company provided drivers. Obviously from an economic standpoint it made sense, the manager likely made 4 times as much as the driver and could do more work in a week. The US managers had no such perk, as I think many Americans enjoy driving, indeed I remember a lawyer telling me his 45 minute commute each way was the best part of his workday. I think though that the younger Borg/smartphone generation will enjoy the opportunity not to be distracted by driving. My preference would be that there are times I would love a robocar, and yet like Will Smith in “I, Robot” would like to have the option to drive myself.

Yes, in the USA, I know billionaires who drive themselves around. And not for the love of driving, but just from habit. It's abundantly clear their time is better spent doing stuff on the ride. This may be changing today because of Uber. It's now become quite common for people who live in SF and many other cities to travel exclusively by Uber/Lyft. Now, if you're a billionaire, it's not a big leap to having a personal chauffeur, which has many advantages over an Uber even though it is more expensive. In some places, especially lower income ones, it is considered very strange for a high income person to drive -- even rude, as they are depriving somebody of a job.

The Uber change tells us these attitudes can change. It is possible a company might start by offering rides only to management, but it will get so cheap that they will consider it for the rank and file for the reasons I outline above. As the cost of the ride gets down below minimum wage, there is certainly no knowledge worker's time that is not vastly more valuable than the cost of the car. Blue collar folks can't as easily do their work in the car, but the truth is there probably are not even many blue collar workers who don't spend some fraction of their day on the computer or phone on company business.

That is an interesting story about the UK. I am interested in whether those cars were parked at the workplace or did the chauffeurs drive them away to somewhere else where parking might have been cheaper? Urban land in UK cities is generally 100 TIMES more expensive than in equivalent US cities, which makes for all sorts of different outcomes. Supermarket aisles are narrower, for example; legroom in theatres is tighter; there are fewer or no familiar (in the US) businesses with low value bulky products (such as "building material recyclers" and car dismantlers). And of course workplace carparks are much more rare. There is a lot more multi-level parking where parking is an essential aspect of the business, for example, at shopping malls - because the cost of building extra levels is less than the cost of surface space.

Interesting article, thanks Brad. I have heard of the idea of employers paying for employees commutes before, as a systemic-level proposal (I think Paolo Pezzotta talks about this) but of course autonomous vehicles increases the scope for this.

One of the mechanisms by which urban property markets respond to, and minimise, congestion delays, is by dispersion of employment. There are numerous input factors that make dispersion a norm, as long as there is market freedom for the dispersion to occur. Higher congestion represents a stronger signal for dispersion than otherwise. Of course there are input factors for concentration as well, such as clustering efficiencies, but as a rule these are outweighed by the dispersion input factors.

Clustering efficiencies on aggregate can often be increased by dispersion in the aggregate, because clustering efficiencies are of multiple types and having "more of them" rather than fewer of them can be optimum. Planners like to think they are creating clustering efficiencies by forcing economic activity into a smaller footprint, but nothing creates clustering efficiencies like the market left to itself. This kind of planning certainly FOREGOES efficiencies of which the planners cannot possibly know.

Seeing the UK has had this kind of urban planning for decades, it is interesting that they have a "productivity gap" for which the most plausible explanation (from Alan W. Evans, Paul Cheshire, and various colleagues) is foregone efficiencies because of such a highly prescriptive planning system.

One of the ironies of compact-city urban planning is that it never results in lower average commute times. Low density US cities might be 5 times further from one fringe to the other, and yet commuters in the "compact city" take just as long or longer on average. I believe the main reason for this is that higher property prices for all kinds of property, in the higher-density cities, actually limit location choices and there are far more compromised location decisions. In contrast, the fact that a US city has 5 times longer distances in the overall, does not mean that people are so stupid as to live a long way away from the job they have; and there are far less "price of housing" barriers to location choice, and employment is highly dispersed anyway. While commuting patterns are random, they tend to be to and from relatively close locations, on average.

Great analysis. I do think there are some rocks left unexplored.

- What if the ads are outside the car, meant for pedestrians? A van with a TV on the side generates far more CPMs driving through NYC than if the TVs were inside. In cities this could easily tip the equation.

- Nationwide average drive commute is 25 minutes one way.[1] Turning that much driving time into passenger time is $$ for Google / Facebook, and other companies that generally make $$ as you spend more time on the internet. Hence their efforts in increasing internet access. I'm not sure what the equation is (eg 1 hour on the internet == $0.X0 for Google), but if it even brings the equation close to breakeven it's good enough for Google.

- Realtime cameras everywhere. This seems at minimum strategically important for a company trying to organize the world's information. It'd likely be the basis for large new platforms as well as improve existing ones like Maps.

1 -

We had a discussion here of moving billboards. They could become a menace, and I see them as something that might get banned if there are too many of them. Today they are only common in very limited areas but with no driver cost that could change.

Yes, the internet advertising companies make money from you while you use your phone in the car. The question is, does that increase your total media exposure time in the whole day? Ie. if you spend 4 hours consuming media now, do you jump to 5 hours adding your car time, or do you just move one of those hours, or is it a mix?

Add new comment