Proof by contradiction: If traffic/speed (aka price) doesn't improve at all, then the demand will not increase and there won't be more cars on the road. Therefore there will be less traffic.
Another thought provoking piece from you. I can't help but point out that I'm sure this level of analysis of transportation networks has most certainly been done, and then, promptly overrulled for some nebulous "political" rationale.
> A town which in the last eight years approved 1,600 new homes and within a week, saw a dozen development proposals filed that put 4,000 new homes in the pipeline
Hypo: part of the answer is that builders are repeat players in a game with politicians and zoning boards, and are worried about being punished in future deals where cities have veto rights for using the builder's remedy too much right now.
I came to the same conclusion. These developers are still participants in the complex dance between loud NIMBYs, zoning commissions, and local politicians. They don't want to ruffle feathers beyond these initial test cases, lest they harm their other projects. These developments don't happen in a vacuum
It doesn’t seem obvious to me that there is capacity to respond elastically to this opportunity. I think each project represents a significant chunk of labor to draft and submit the plans, and I suspect you need to have capital committed if not own the land?
The simple answer is "induced demand" is a sloppy concept used by anti-car activists who are more motivated by stopping any and all car infrastructure projects rather than seeking the truth. Here's a good Motte post about it ( https://www.reddit.com/r/TheMotte/comments/on89vw/culture_war_roundup_for_the_week_of_july_19_2021/h63ix39/ ). Notably, they never make the same argument against expanding transit or bike infrastructure just because the new capacity will just be used up. This is similar to the notion that new construction raises rents which implies new buildings would just somehow generate demand out of nowhere (that exceeds supply), which you and Matthew Yglesias point out is complete nonsense and would imply the existence of the economics version of the perpetual motion machine.
Similarly, if building new roads generates economic demand out of thin air, then cities could just get rich quick by just building more and more roads. Unfortunately, they can't, and for proof consider the 20-lane highway they built in Myanmar that is completely empty. This is because demand doesn't come from nowhere, but rather comes from people who wanted to use the road but previously couldn't, and this is an actual economics concept known as _latent demand_. Notably, latent demand is less than proportional to previous demand, so there is indeed a limit where the majority of demand can be satisfied and building any new capacity brings increasingly marginal returns, if not just being outright wasteful.
I have yet to see a steelman of "induced demand" that makes sense. Especially when they keep making blatantly false statements like "the Katy freeway is 26 lanes wide" (no, it literally isn't) or fail to consider other causes of congestion like a growing population from immigration.
I would note that a huge chunk of DC's transit policy comes down to managing the challenges of fare evasion.
Why are we making buses free? Because so many people evade paying already that the total incremental cost is shockingly low, and it means that maybe more people will take the bus. We already made it free for a while during the pandemic (out of a desire to socially distance drivers from passengers, they boarded through the rear doors without fairboxes), so we know it has that effect.
Why are so many people jumping turnstiles in the Metro (subway)?
First, because a huge chunk of the fare jumpers at rush hour are kids, who get free transit to and from school on special metro cards that they lose all the time, and they don't want to be late for class.
Second, because DC has a distance-and-peak/non-peak rate system for the Metro subway, it's easy to accidentally not have enough balance on your metro card to exit -- and for reasons passing understanding, the systems inside the turnstile to add value don't take credit cards; this means basically everyone in DC has at some point had to jump a turnstile (e.g., I have, with permission from the station operator).
And third, there are still plenty of folks jumping because they don't want to pay -- but it's important to subtract the first two chunks from the estimate.
(Why not make the Metrorail free? Because it would require VA and MD's cooperation, unlike the buses that are more regionalized, and because we like making tourists pay for Metro).
A lot of the induced demand is not from new driving but instead from people switching off of other roads.
As a thought experiment suppose there were 10 parallel highways and one of them was doubled in size. This would only have a 10% increase in total capacity which could easily be unnoticeable (particularly with a small amount of increased demand)
I think the problem people ascribe to induced demand is really more about the highways that are expanded not actually being the relevant bottlenecks. They just end up being parking lots where people wait their turn to enter the city, whose congested roads are the actual bottleneck. So expanding the highway just means a bigger waiting area, not faster throughput.
My explanation for induced demand was going to be that the level of acceptable congestion always hovers right around the equivalent of driving down a non-highway street with stoplights (hence unsatisfied demand doesn't lead to "slightly worse" congestion on freeways, it just distributes it to alternate routes), but I think your explanation makes more sense. It would be interesting to test it; check traffic further back into the suburbs and see if widening the freeway moves the start of the slowdown closer to downtown, even though it doesn't change speeds near downtown.
> The highway is equally terrible now as it was before, yet twice as many people choose to use it simply because it is bigger? Something has gone very wrong somewhere.
1. If you upgrade the highway capacity without upgrading the number of exits or their capacity, then the marginal value of the upgraded highway will remain very low, as people will simply be stuck in line for the exits instead being stuck on the highway itself.
2. Once the highway is upgraded people start doing more trips or switching their trip times closer to rush hour, up until a new equilibrium is reached. Obviously this cannot last forever and at some point you *can* improve traffic via adding more highways.
3. Better flowing traffic induces people to purchase a car (instead of using public transit) and once they've purchased one they'll likely stick with driving long past the point where it still makes sense. Though this part only applies to cities where many people don't have a car in the first place, such as NYC.
4. I suspect part of the reason why "induced demand" is over-rated is that highway upgrades often coincide with an increase in the number of people living in the city, so what looks like "induced demand" is just a natural increase in traffic resulting from a higher population.
> If you think that such trips have negative externalities, we have tolls and gasoline taxes available to fix that
Unfortunately no city in the world (AFAIK) implements tolls correctly. Ideally you want dynamic tolls with no upper pricing bound where it's *extremely* expensive to drive at rush hour, but free to drive in the middle of the night, so that there's *never* any serious traffic jams. But even London has a pretty low cap on their "congestion charge" (~$20), so their roads still have traffic jams, which is particularly bad for their bus lines, as they don't always have dedicated lanes.
I predict that living in a city with uncapped dynamic tolls would be a fantastic experience, as you could always predict how long it would take you to get from A to B, either by bus or by car. And it would reduce the need for expensive underground transportation which is not a viable solution for many low-density locations.
> The danger with buses, as I see it, is providing coverage that is too good where you don’t need it. When I lived in Warwick, there was a bus that could take us into New York City that came every 30 minutes or so. This was highly valuable. However, except at rush hour, the bus was mostly empty. If that bus had mostly come every 60 minutes instead, how much of the value would have been retained?
I imagine the optimal solution would be two have *two* buses. One to take you to the closest "transit center" and one from the transit center to NYC. If both buses run every 15 mins, everyone would be better off and it will likely be cheaper for the MTA as well.
I think it's important to note that highway expansion usually happens when there is already a ton of traffic. Current drivers complain, the mayor/governor spends a ton of money to build another lane promising that it will reduce traffic, and a year later those same drivers are often in more traffic than they were previously.
It has induced new trips from new drivers, who otherwise would have taken public transit or not made those trips at all, but the people who the benefits were sold to usually don't see any.
I think this a key distinction between road and public transit capital projects. If my bus route is replaced with a subway I am better off indefinitely with a faster more reliable commute. Whereas if my commute freeway is widened mostly other people benefit to the extent that they make trips when they otherwise would not have but the freeway remains congested.
Ironically discourse tends to justify such projects on reverse grounds. "New riders" for transit verse faster travel times ("congestion relief") for freeway expansion.
I think it's also relevant that public transit ridership is a "passive" activity, where you can do other potentially productive things during that time. Whereas driving is active and no productive activity can occur.
Why hasn't GPS solved the traffic problem? I have a commute of about five miles, with three legitimate ways home. The GPS almost always takes me on a highway that is moving at about 10 MPH. The reason is that this is faster and more direct than taking the city streets. So might I suggest that people congest the highways because even slow they are the fastest way to get from point A to point B.
Highway architecture plays a part in backups as well. Recently, a sweeping curve replace a hard left turn from one highway to another (with a traffic light at the intersection. Instead of having to slow down to make the 90 degree turn, the sweep allows one to go through at full speed. Traffic no longer backs up onto the highway from the exit, since more cars get through on a light cycle. (This is the I-90 I-291 interchange).
my understanding is that gps has allowed more people to use neighborhood roads, but that just makes traffic flow and expand to fill potential space (and smart locals were already using neighborhood roads before)
In 2020-2021, I was a tenant in a building in California that used RealPage. When I shopping for housing in 2020, the building advertised a "concession" of 6 weeks free rent — rather than lowering the price outright. When my lease was up for renewal in 2021, they asked for an increase of 10+%.
In practical terms, this meant a rent increase of 20+%.
I pointed out to building management that this was absurd, that it put the price above market when compared to nearby units, and that the number of units listed as available suggested it'd be in their interest to retain me as a tenant.
The on-site building staff were unable to negotiate. Negotiations were handled by specialists in a call center out of state.
So I called, and repeated my pitch.
I was offered two options:
• Accept the offer.
• Move to a smaller, cheaper unit.
I proposed two alternatives:
• Renew for 12 months at my current rate (which without the concession would still be an effective 10+% increase).
• Make a 18/24/36-month commitment in exchange for ~5% discount over the new rate (which without the concession would be an effective 15+% increase).
I was told they could only offer me the options presented to them by the computer.
I tried to haggle a little too see if there was any wiggle room. How about waiving the pet rent (~$100/month)? How about a storage cage in the garage (~$50/month)? How about a couple free guest parking passes per month (~$30 value)?
Nope. Couldn't even budge by, essentially, 0.5% of the new price.
So I go back to the on-site staff and let them know we're moving out. I tried to probe a bit: given the cost of finding a new tenant, how did this make any sense?
They admitted that it didn't. The building was owned by an insurance company. As far as they could tell, the company cared about maximizing the average rent price, not occupancy or totally monthly revenue.
They suspected the owner would have in theory been interested in some of the counter-offers I'd made, but confirmed there was no mechanism to actually consider counter-offers. You accept or move out. You can't talk to anyone empowered to use logic.
After we moved out, when the unit was eventually listed, it had new flooring. It was listed for more than my rent price, but less than the asking renewal price. It sat vacant for months and the price slowly dropped. I wasn't paying close enough attention to see the final price, but it was less than what I would have negotiated to.
I still can't find a rational explanation. I would have renewed at a higher price, and the unit wouldn't have sat vacant. They left a lot of money on the table.
Two basic possibilities. Either they have stupid KPIs, as you suggest, and are not optimizing for profits, or they are making a gamble that some people will suck it up - it is common to run the 'charge renewal a lot' gambit when you think they might suck it up and pay it. It's just that in this case they're wrong. But the hope was that by tying their hands - being unable to negotiate - they'd get you to cave where you wouldn't have to a human. Didn't work out.
Re Induced demand for roads: Yeah I'm going to say the demand is very high, so until you have enough roads such that they are all always running at full speed, there is more demand. I use to commute every week day into Buffalo, (known for having much smaller traffic jams than other areas), I would still plan my day (when I could) to avoid the rush hours on the highways. I would drive in earlier or later in the morning and the same for the evening. Given that there are probably many people adjusting their driving habits to avoid the congestion, well that's the excess demand. And we all have different levels of how much traffic is too much, and so new traffic lanes fill up, till they reach that point where enough people are adjusting their driving habits to account for the congestion.
Proof by contradiction: If traffic/speed (aka price) doesn't improve at all, then the demand will not increase and there won't be more cars on the road. Therefore there will be less traffic.
I know, right?
Another thought provoking piece from you. I can't help but point out that I'm sure this level of analysis of transportation networks has most certainly been done, and then, promptly overrulled for some nebulous "political" rationale.
> What struck me most about Santa Monica was that there wasn’t way more new construction approved on the first day.
Perhaps this hinges on what you intended by “way more”, but I think we did see way more construction in Santa Monica, see this deep dive: https://darrellowens.substack.com/p/ca-cities-to-lose-all-zoning-powers
> A town which in the last eight years approved 1,600 new homes and within a week, saw a dozen development proposals filed that put 4,000 new homes in the pipeline
Sure, but given the unique opportunity, the question of why not more remains super interesting.
Hypo: part of the answer is that builders are repeat players in a game with politicians and zoning boards, and are worried about being punished in future deals where cities have veto rights for using the builder's remedy too much right now.
I came to the same conclusion. These developers are still participants in the complex dance between loud NIMBYs, zoning commissions, and local politicians. They don't want to ruffle feathers beyond these initial test cases, lest they harm their other projects. These developments don't happen in a vacuum
It doesn’t seem obvious to me that there is capacity to respond elastically to this opportunity. I think each project represents a significant chunk of labor to draft and submit the plans, and I suspect you need to have capital committed if not own the land?
The simple answer is "induced demand" is a sloppy concept used by anti-car activists who are more motivated by stopping any and all car infrastructure projects rather than seeking the truth. Here's a good Motte post about it ( https://www.reddit.com/r/TheMotte/comments/on89vw/culture_war_roundup_for_the_week_of_july_19_2021/h63ix39/ ). Notably, they never make the same argument against expanding transit or bike infrastructure just because the new capacity will just be used up. This is similar to the notion that new construction raises rents which implies new buildings would just somehow generate demand out of nowhere (that exceeds supply), which you and Matthew Yglesias point out is complete nonsense and would imply the existence of the economics version of the perpetual motion machine.
Similarly, if building new roads generates economic demand out of thin air, then cities could just get rich quick by just building more and more roads. Unfortunately, they can't, and for proof consider the 20-lane highway they built in Myanmar that is completely empty. This is because demand doesn't come from nowhere, but rather comes from people who wanted to use the road but previously couldn't, and this is an actual economics concept known as _latent demand_. Notably, latent demand is less than proportional to previous demand, so there is indeed a limit where the majority of demand can be satisfied and building any new capacity brings increasingly marginal returns, if not just being outright wasteful.
I have yet to see a steelman of "induced demand" that makes sense. Especially when they keep making blatantly false statements like "the Katy freeway is 26 lanes wide" (no, it literally isn't) or fail to consider other causes of congestion like a growing population from immigration.
I would note that a huge chunk of DC's transit policy comes down to managing the challenges of fare evasion.
Why are we making buses free? Because so many people evade paying already that the total incremental cost is shockingly low, and it means that maybe more people will take the bus. We already made it free for a while during the pandemic (out of a desire to socially distance drivers from passengers, they boarded through the rear doors without fairboxes), so we know it has that effect.
Why are so many people jumping turnstiles in the Metro (subway)?
First, because a huge chunk of the fare jumpers at rush hour are kids, who get free transit to and from school on special metro cards that they lose all the time, and they don't want to be late for class.
Second, because DC has a distance-and-peak/non-peak rate system for the Metro subway, it's easy to accidentally not have enough balance on your metro card to exit -- and for reasons passing understanding, the systems inside the turnstile to add value don't take credit cards; this means basically everyone in DC has at some point had to jump a turnstile (e.g., I have, with permission from the station operator).
And third, there are still plenty of folks jumping because they don't want to pay -- but it's important to subtract the first two chunks from the estimate.
(Why not make the Metrorail free? Because it would require VA and MD's cooperation, unlike the buses that are more regionalized, and because we like making tourists pay for Metro).
A lot of the induced demand is not from new driving but instead from people switching off of other roads.
As a thought experiment suppose there were 10 parallel highways and one of them was doubled in size. This would only have a 10% increase in total capacity which could easily be unnoticeable (particularly with a small amount of increased demand)
I think the problem people ascribe to induced demand is really more about the highways that are expanded not actually being the relevant bottlenecks. They just end up being parking lots where people wait their turn to enter the city, whose congested roads are the actual bottleneck. So expanding the highway just means a bigger waiting area, not faster throughput.
My explanation for induced demand was going to be that the level of acceptable congestion always hovers right around the equivalent of driving down a non-highway street with stoplights (hence unsatisfied demand doesn't lead to "slightly worse" congestion on freeways, it just distributes it to alternate routes), but I think your explanation makes more sense. It would be interesting to test it; check traffic further back into the suburbs and see if widening the freeway moves the start of the slowdown closer to downtown, even though it doesn't change speeds near downtown.
> The highway is equally terrible now as it was before, yet twice as many people choose to use it simply because it is bigger? Something has gone very wrong somewhere.
1. If you upgrade the highway capacity without upgrading the number of exits or their capacity, then the marginal value of the upgraded highway will remain very low, as people will simply be stuck in line for the exits instead being stuck on the highway itself.
2. Once the highway is upgraded people start doing more trips or switching their trip times closer to rush hour, up until a new equilibrium is reached. Obviously this cannot last forever and at some point you *can* improve traffic via adding more highways.
3. Better flowing traffic induces people to purchase a car (instead of using public transit) and once they've purchased one they'll likely stick with driving long past the point where it still makes sense. Though this part only applies to cities where many people don't have a car in the first place, such as NYC.
4. I suspect part of the reason why "induced demand" is over-rated is that highway upgrades often coincide with an increase in the number of people living in the city, so what looks like "induced demand" is just a natural increase in traffic resulting from a higher population.
> If you think that such trips have negative externalities, we have tolls and gasoline taxes available to fix that
Unfortunately no city in the world (AFAIK) implements tolls correctly. Ideally you want dynamic tolls with no upper pricing bound where it's *extremely* expensive to drive at rush hour, but free to drive in the middle of the night, so that there's *never* any serious traffic jams. But even London has a pretty low cap on their "congestion charge" (~$20), so their roads still have traffic jams, which is particularly bad for their bus lines, as they don't always have dedicated lanes.
I predict that living in a city with uncapped dynamic tolls would be a fantastic experience, as you could always predict how long it would take you to get from A to B, either by bus or by car. And it would reduce the need for expensive underground transportation which is not a viable solution for many low-density locations.
> The danger with buses, as I see it, is providing coverage that is too good where you don’t need it. When I lived in Warwick, there was a bus that could take us into New York City that came every 30 minutes or so. This was highly valuable. However, except at rush hour, the bus was mostly empty. If that bus had mostly come every 60 minutes instead, how much of the value would have been retained?
I imagine the optimal solution would be two have *two* buses. One to take you to the closest "transit center" and one from the transit center to NYC. If both buses run every 15 mins, everyone would be better off and it will likely be cheaper for the MTA as well.
I think it's important to note that highway expansion usually happens when there is already a ton of traffic. Current drivers complain, the mayor/governor spends a ton of money to build another lane promising that it will reduce traffic, and a year later those same drivers are often in more traffic than they were previously.
It has induced new trips from new drivers, who otherwise would have taken public transit or not made those trips at all, but the people who the benefits were sold to usually don't see any.
I think this a key distinction between road and public transit capital projects. If my bus route is replaced with a subway I am better off indefinitely with a faster more reliable commute. Whereas if my commute freeway is widened mostly other people benefit to the extent that they make trips when they otherwise would not have but the freeway remains congested.
Ironically discourse tends to justify such projects on reverse grounds. "New riders" for transit verse faster travel times ("congestion relief") for freeway expansion.
I think it's also relevant that public transit ridership is a "passive" activity, where you can do other potentially productive things during that time. Whereas driving is active and no productive activity can occur.
Why hasn't GPS solved the traffic problem? I have a commute of about five miles, with three legitimate ways home. The GPS almost always takes me on a highway that is moving at about 10 MPH. The reason is that this is faster and more direct than taking the city streets. So might I suggest that people congest the highways because even slow they are the fastest way to get from point A to point B.
Highway architecture plays a part in backups as well. Recently, a sweeping curve replace a hard left turn from one highway to another (with a traffic light at the intersection. Instead of having to slow down to make the 90 degree turn, the sweep allows one to go through at full speed. Traffic no longer backs up onto the highway from the exit, since more cars get through on a light cycle. (This is the I-90 I-291 interchange).
my understanding is that gps has allowed more people to use neighborhood roads, but that just makes traffic flow and expand to fill potential space (and smart locals were already using neighborhood roads before)
RealPage annecdote:
In 2020-2021, I was a tenant in a building in California that used RealPage. When I shopping for housing in 2020, the building advertised a "concession" of 6 weeks free rent — rather than lowering the price outright. When my lease was up for renewal in 2021, they asked for an increase of 10+%.
In practical terms, this meant a rent increase of 20+%.
I pointed out to building management that this was absurd, that it put the price above market when compared to nearby units, and that the number of units listed as available suggested it'd be in their interest to retain me as a tenant.
The on-site building staff were unable to negotiate. Negotiations were handled by specialists in a call center out of state.
So I called, and repeated my pitch.
I was offered two options:
• Accept the offer.
• Move to a smaller, cheaper unit.
I proposed two alternatives:
• Renew for 12 months at my current rate (which without the concession would still be an effective 10+% increase).
• Make a 18/24/36-month commitment in exchange for ~5% discount over the new rate (which without the concession would be an effective 15+% increase).
I was told they could only offer me the options presented to them by the computer.
I tried to haggle a little too see if there was any wiggle room. How about waiving the pet rent (~$100/month)? How about a storage cage in the garage (~$50/month)? How about a couple free guest parking passes per month (~$30 value)?
Nope. Couldn't even budge by, essentially, 0.5% of the new price.
So I go back to the on-site staff and let them know we're moving out. I tried to probe a bit: given the cost of finding a new tenant, how did this make any sense?
They admitted that it didn't. The building was owned by an insurance company. As far as they could tell, the company cared about maximizing the average rent price, not occupancy or totally monthly revenue.
They suspected the owner would have in theory been interested in some of the counter-offers I'd made, but confirmed there was no mechanism to actually consider counter-offers. You accept or move out. You can't talk to anyone empowered to use logic.
After we moved out, when the unit was eventually listed, it had new flooring. It was listed for more than my rent price, but less than the asking renewal price. It sat vacant for months and the price slowly dropped. I wasn't paying close enough attention to see the final price, but it was less than what I would have negotiated to.
I still can't find a rational explanation. I would have renewed at a higher price, and the unit wouldn't have sat vacant. They left a lot of money on the table.
Two basic possibilities. Either they have stupid KPIs, as you suggest, and are not optimizing for profits, or they are making a gamble that some people will suck it up - it is common to run the 'charge renewal a lot' gambit when you think they might suck it up and pay it. It's just that in this case they're wrong. But the hope was that by tying their hands - being unable to negotiate - they'd get you to cave where you wouldn't have to a human. Didn't work out.
Re Induced demand for roads: Yeah I'm going to say the demand is very high, so until you have enough roads such that they are all always running at full speed, there is more demand. I use to commute every week day into Buffalo, (known for having much smaller traffic jams than other areas), I would still plan my day (when I could) to avoid the rush hours on the highways. I would drive in earlier or later in the morning and the same for the evening. Given that there are probably many people adjusting their driving habits to avoid the congestion, well that's the excess demand. And we all have different levels of how much traffic is too much, and so new traffic lanes fill up, till they reach that point where enough people are adjusting their driving habits to account for the congestion.