That's right, but if demand is perfectly elastic, there's no welfare gains from these extra commuters. All of them are just barely indifferent between the benefit they get from commuting and the time cost they have to pay.
I don’t buy this- building roads leads to lots more people doing whatever they do- all economic activity is kind of at the margin, but more is better. And the number of people goes up with the square of the radius served my the new roads. https://chatgpt.com/share/6a8dcdec-5878-83ea-af75-edcf3efcb7cc
You're right except when demand is perfectly elastic. Then everyone has the same maximum tolerable commute, and supply shifts induce more entry until everyone is paying that same maximum tolerable commute again.
Even if demand is not _perfectly_ elastic, when it's super flat most of the new entrants get almost nothing out of commuting and the incumbents don't benefit either because all the new entrants bid up the time cost very close to what it was before.
I was thinking along the same lines. Specifically, does this mean that in cities with better public transit (i.e. where there is higher elasticity of demand for highways because commuters have options), adding lanes to the highway is more impactful?
Seems like an interesting way to say that if you want to increase the value/benefit of your existing lanes, create public transport!
Hmmm, I think you skipped too quickly over the part where you investigated the existing research. ChatGPT's analysis:
"Adding lanes generally does increase capacity, so holding everything else fixed it reduces congestion. Induced demand means some of that improvement gets eaten away over time; it does not mean the capacity increase literally has no effect in every case."
So as far as I can tell, Joe is correct here. Lanes and housing work the same way. Adding more lanes generally reduces congestion, and adding more housing generally reduces housing prices. Neither one of them is as effective as a linear model would predict, because there are second-order effects.
There is the empirical claim that demand for highways is highly elastic. That may or may not be true, and it's certainly not true in all cases.
But the real heart of the essay is the conditional claim: _If_ highways and houses have elastic demand, then building highways is useless but building houses is still good.
This difference between houses and highways comes down to differences in they are priced. The price for each driver entering a highway is the average cost of congestion in minutes. The price for a renter renting a house is the marginal cost of construction in dollars.
The fact that each driver pays the average cost of congestion means that total "revenue" (price*quantity but here price is measured in time spent) always equals total cost.
This isn't the case with marginal priced goods. The market price equals the cost of producing the last unit, while earlier units generally cost less to produce. When supply shifts right, those earlier units become cheaper to produce. So even if the final market price and the cost of producing the last unit is the same, more welfare is generated on those earlier units.
But with average cost pricing, the surplus generated on the first few units that cost less than average to make is always exactly canceled out by losses generated on the last units that cost more than average.
So then the only source of surplus is from a lower price, but elastic demand prevents that.
I believe that conditional claim is wrong as stated. If highways have *infinitely* elastic demand, then building highways is useless. With any finitely elastic demand, congestion will fall.
yes that's true, but demand elasticity is finite but very high then still the welfare gains from building highways are tiny. But that's not true for building houses!
Okay so we are just bickering about small versus zero. Fair.
Maybe off topic at this point. But it's kind of funny how these two issues are tied together, although sort of backwards rather than by analogy.
Ie if your ideal way of life is to have a 20 minute commute by car and a large house, like you say, it's not very effective to advocate for more lanes whenever traffic gets too high. It's much more effective to advocate for restricting new housing, before traffic gets too bad in the first place! In a sense, the more correct your argument is, the more that car-loving NIMBYs should oppose new housing in the first place.
I love this post - I’m using it in Micro 101 this semester when I discuss elasticity. Hopefully it will bring to life a term that usually causes my students to immediately fall asleep! 🙏
But building extra lanes of highway isn't useless even if it doesn't reduce congestion. If you add a third land, you have 1/3rd more people enjoying all the benefits of going somewhere and doing something. The base reason for adding an additional lane isn't to 'reduce congestion'; the base reason for adding an additional lane is to allow people to go places they want to go, and to do things they want to do; resulting in a large net increase in welfare. Maybe it will even increase housing prices-- due to the network effect you mentioned.
This is incorrect when demand is perfectly elastic, and approximately wrong when demand is highly elastic and the shift in supply is small.
Under these conditions, all of the new drivers are either exactly indifferent or close to it. If you look at the gray triangle on the highway graph you'll see that the difference between the benefit that new drivers get and the cost they pay is very small. When the demand curve is flat, it's zero.
So all those extra people who get to commute at the same level of congestion are not actually getting any extra welfare out of it.
The grey triangle is based on AN ASSUMPTION, on a MODEL. If planners added a third lane and it immediately fills up, to the point where the new congestion is equal to the old congestion, that implies there was a very large pool of potential drivers, thousands, if not tens of thousands, who are completely willing, indeed, apparently eager, to commute at those levels of congestion. They were frustrated in that desire because when they tried to travel at those levels of congestion, congestion increased, and taking the trip at a higher level of congestion was a deal they were unwilling to take. Congestion isn’t a linear relationship. Indeed it is very non-linear, most especially at higher levels of congestion. At a high level of existing congestion, if you add a few percent more cars you can get big discontinuities in travel time, including dead stop. Since congestion is nonlinear, indeed very non-linear at higher levels of traffic, the penalty the drivers have to pay is very nonlinear as well. And that means you can’t model the pay off to the drivers using a linear equation.
Indeed many models predict that housing prices rise as housing supply expands.
This is an artifact of what is take as "price" of housing. The result is a combination of composition effects (if you eyedropper out a bit of liberalization, the marginal unit to be built with be very high cost) plus non-hedonic price comparisons.
This is why economics is called the dismal science. In real life, charts help nothing. The only way to find out is to try it and see if it works. I think it was St. Milton who said that economic forecasting was impossible because the most important parameter, erratic and unstable politicans, couldn't be factored in. Scott Adams who was an economic forecaster before he met Dilbert, was likewise negative about models.
Even if adding lanes doesn’t reduce congestion, it increases the number of people who can travel on the congested roads.
That's right, but if demand is perfectly elastic, there's no welfare gains from these extra commuters. All of them are just barely indifferent between the benefit they get from commuting and the time cost they have to pay.
I don’t buy this- building roads leads to lots more people doing whatever they do- all economic activity is kind of at the margin, but more is better. And the number of people goes up with the square of the radius served my the new roads. https://chatgpt.com/share/6a8dcdec-5878-83ea-af75-edcf3efcb7cc
I thought only marginal commuters would be barely indifferent.
You're right except when demand is perfectly elastic. Then everyone has the same maximum tolerable commute, and supply shifts induce more entry until everyone is paying that same maximum tolerable commute again.
Even if demand is not _perfectly_ elastic, when it's super flat most of the new entrants get almost nothing out of commuting and the incumbents don't benefit either because all the new entrants bid up the time cost very close to what it was before.
So if we manage to hasten the variably-tolled future, does that mean that building more lanes will become important again?
I was thinking along the same lines. Specifically, does this mean that in cities with better public transit (i.e. where there is higher elasticity of demand for highways because commuters have options), adding lanes to the highway is more impactful?
Seems like an interesting way to say that if you want to increase the value/benefit of your existing lanes, create public transport!
Hmmm, I think you skipped too quickly over the part where you investigated the existing research. ChatGPT's analysis:
"Adding lanes generally does increase capacity, so holding everything else fixed it reduces congestion. Induced demand means some of that improvement gets eaten away over time; it does not mean the capacity increase literally has no effect in every case."
So as far as I can tell, Joe is correct here. Lanes and housing work the same way. Adding more lanes generally reduces congestion, and adding more housing generally reduces housing prices. Neither one of them is as effective as a linear model would predict, because there are second-order effects.
I think you're confused in your response here.
There is the empirical claim that demand for highways is highly elastic. That may or may not be true, and it's certainly not true in all cases.
But the real heart of the essay is the conditional claim: _If_ highways and houses have elastic demand, then building highways is useless but building houses is still good.
This difference between houses and highways comes down to differences in they are priced. The price for each driver entering a highway is the average cost of congestion in minutes. The price for a renter renting a house is the marginal cost of construction in dollars.
The fact that each driver pays the average cost of congestion means that total "revenue" (price*quantity but here price is measured in time spent) always equals total cost.
This isn't the case with marginal priced goods. The market price equals the cost of producing the last unit, while earlier units generally cost less to produce. When supply shifts right, those earlier units become cheaper to produce. So even if the final market price and the cost of producing the last unit is the same, more welfare is generated on those earlier units.
But with average cost pricing, the surplus generated on the first few units that cost less than average to make is always exactly canceled out by losses generated on the last units that cost more than average.
So then the only source of surplus is from a lower price, but elastic demand prevents that.
I believe that conditional claim is wrong as stated. If highways have *infinitely* elastic demand, then building highways is useless. With any finitely elastic demand, congestion will fall.
yes that's true, but demand elasticity is finite but very high then still the welfare gains from building highways are tiny. But that's not true for building houses!
Okay so we are just bickering about small versus zero. Fair.
Maybe off topic at this point. But it's kind of funny how these two issues are tied together, although sort of backwards rather than by analogy.
Ie if your ideal way of life is to have a 20 minute commute by car and a large house, like you say, it's not very effective to advocate for more lanes whenever traffic gets too high. It's much more effective to advocate for restricting new housing, before traffic gets too bad in the first place! In a sense, the more correct your argument is, the more that car-loving NIMBYs should oppose new housing in the first place.
I love this post - I’m using it in Micro 101 this semester when I discuss elasticity. Hopefully it will bring to life a term that usually causes my students to immediately fall asleep! 🙏
Very nice point
But building extra lanes of highway isn't useless even if it doesn't reduce congestion. If you add a third land, you have 1/3rd more people enjoying all the benefits of going somewhere and doing something. The base reason for adding an additional lane isn't to 'reduce congestion'; the base reason for adding an additional lane is to allow people to go places they want to go, and to do things they want to do; resulting in a large net increase in welfare. Maybe it will even increase housing prices-- due to the network effect you mentioned.
This is incorrect when demand is perfectly elastic, and approximately wrong when demand is highly elastic and the shift in supply is small.
Under these conditions, all of the new drivers are either exactly indifferent or close to it. If you look at the gray triangle on the highway graph you'll see that the difference between the benefit that new drivers get and the cost they pay is very small. When the demand curve is flat, it's zero.
So all those extra people who get to commute at the same level of congestion are not actually getting any extra welfare out of it.
The grey triangle is based on AN ASSUMPTION, on a MODEL. If planners added a third lane and it immediately fills up, to the point where the new congestion is equal to the old congestion, that implies there was a very large pool of potential drivers, thousands, if not tens of thousands, who are completely willing, indeed, apparently eager, to commute at those levels of congestion. They were frustrated in that desire because when they tried to travel at those levels of congestion, congestion increased, and taking the trip at a higher level of congestion was a deal they were unwilling to take. Congestion isn’t a linear relationship. Indeed it is very non-linear, most especially at higher levels of congestion. At a high level of existing congestion, if you add a few percent more cars you can get big discontinuities in travel time, including dead stop. Since congestion is nonlinear, indeed very non-linear at higher levels of traffic, the penalty the drivers have to pay is very nonlinear as well. And that means you can’t model the pay off to the drivers using a linear equation.
Indeed many models predict that housing prices rise as housing supply expands.
This is an artifact of what is take as "price" of housing. The result is a combination of composition effects (if you eyedropper out a bit of liberalization, the marginal unit to be built with be very high cost) plus non-hedonic price comparisons.
This is why economics is called the dismal science. In real life, charts help nothing. The only way to find out is to try it and see if it works. I think it was St. Milton who said that economic forecasting was impossible because the most important parameter, erratic and unstable politicans, couldn't be factored in. Scott Adams who was an economic forecaster before he met Dilbert, was likewise negative about models.