The economic impact of the strike will be...
October 6, 2015 8:46 AM Subscribe
...a net positive. We often hear about how much strike will cost the economy, however new research suggests the London underground train strike of early 2014 may have positive effects as it forced commuters to find different routes to work, many of which they stuck with. Researchers were able to access 200 million data points from the Oyster Card system that allows access to the London transport system to conclude about 5% of travellers switched their route after finding a new route during the strike, suggesting a lack of experimentation concerning the available options. Short paper here.
This might be from people realizing that stops that look far apart on the underground map are actually quite close when forced to walk between them.
posted by srboisvert at 9:06 AM on October 6, 2015 [2 favorites]
posted by srboisvert at 9:06 AM on October 6, 2015 [2 favorites]
You guys buy this?
Maybe brits are a little set in their ways? Who doesn't optimize their commute and continue to refine it?
posted by Reasonably Everything Happens at 9:13 AM on October 6, 2015 [2 favorites]
Maybe brits are a little set in their ways? Who doesn't optimize their commute and continue to refine it?
posted by Reasonably Everything Happens at 9:13 AM on October 6, 2015 [2 favorites]
I can try different routes now because I don't get in trouble if I get to work late. But for other jobs, it isn't worth risking that a different method will take a lot longer on the chance it will be shorter -- I know that my main route takes X time and I'm ok with it.
posted by jeather at 9:14 AM on October 6, 2015 [1 favorite]
posted by jeather at 9:14 AM on October 6, 2015 [1 favorite]
Or it could be completely wrong.
What I don't see is any evidence that they checked that other routes changed state during the strike. If, because of a tube strike, more buses, regular trains, and taxis were on the street, that "more efficient route" may disappear the moment the tube starts running and those extra vehicles are pulled back off the street.
Plus, the assumption that the "benefit" will last for a long time is fundamentally flaw. While tube schedules change frequently, the trains run along the tracks. Other than station closures, they're not going to magically reroute themselves, and the few that even have the ability to change routes change *very* infrequently because of the knock on effect. Buses, however, change both routes and schedules. Your newly more efficient route could well disappear in the next week when the bus changes route because of road construction, or load changes, and so forth.
IOW, this is a vastly simplified case built on limited data set -- the few tube strikes that have occurred since Oyster card data was put out, and published with no peer review.
posted by eriko at 9:14 AM on October 6, 2015 [2 favorites]
What I don't see is any evidence that they checked that other routes changed state during the strike. If, because of a tube strike, more buses, regular trains, and taxis were on the street, that "more efficient route" may disappear the moment the tube starts running and those extra vehicles are pulled back off the street.
Plus, the assumption that the "benefit" will last for a long time is fundamentally flaw. While tube schedules change frequently, the trains run along the tracks. Other than station closures, they're not going to magically reroute themselves, and the few that even have the ability to change routes change *very* infrequently because of the knock on effect. Buses, however, change both routes and schedules. Your newly more efficient route could well disappear in the next week when the bus changes route because of road construction, or load changes, and so forth.
IOW, this is a vastly simplified case built on limited data set -- the few tube strikes that have occurred since Oyster card data was put out, and published with no peer review.
posted by eriko at 9:14 AM on October 6, 2015 [2 favorites]
They should could continue to force people to examine their commute options the way the New York MTA does: randomly take routes out of operation either for certain hours or just eliminate the route entirely. Breaking everyone's commute every 6 months or so guarantees that nobody gets settled. Hopefully Londoners are like New Yorkers in that everyone has some idea of how you should get from place to place that's better than your idea.
posted by Sunburnt at 9:27 AM on October 6, 2015
posted by Sunburnt at 9:27 AM on October 6, 2015
IOW, this is a vastly simplified case built on limited data set -- the few tube strikes that have occurred since Oyster card data was put out, and published with no peer review.
True enough, but the finding does seem to make intuitive sense. It's basically describing simulated annealing in an optimization problem being performed by humans. Adding a little noise to the system helps bounce a small population of people who were stuck in a local minimum (a route that was satisfactory) over to either a new better local minimum (an even more satisfactory route) or the global optimum (the actual objective best possible route).
Either that, or if you force people out of their routine and make them think about planning their route once it a while, some of them will notice that they can come up with better solutions.
posted by figurant at 9:53 AM on October 6, 2015 [2 favorites]
True enough, but the finding does seem to make intuitive sense. It's basically describing simulated annealing in an optimization problem being performed by humans. Adding a little noise to the system helps bounce a small population of people who were stuck in a local minimum (a route that was satisfactory) over to either a new better local minimum (an even more satisfactory route) or the global optimum (the actual objective best possible route).
Either that, or if you force people out of their routine and make them think about planning their route once it a while, some of them will notice that they can come up with better solutions.
posted by figurant at 9:53 AM on October 6, 2015 [2 favorites]
Who doesn't optimize their commute and continue to refine it?
Optimize? Everyone, to an extent. Continue to refine it? Nobody, to an extent.
The linked paper posits two (non-exclusionary) scenarios:
1. People optimized their routes until it the "costs" of increasing that optimization made it less worth seeking out maximal optimization.
2. People found a route that they were satisfied with and didn't give it much more thought.
Some combination of those two explanations strikes me as a good fit for how most people actually approach their work commute. I have a few routes I can take to-and-from the office, but I generally don't give them much thought because it is goddamn 7 in the morning and fuck this traffic and I'm erasing this whole experience from my memory as soon as I get to my destination for the sake of my sanity.
posted by Panjandrum at 9:55 AM on October 6, 2015
Optimize? Everyone, to an extent. Continue to refine it? Nobody, to an extent.
The linked paper posits two (non-exclusionary) scenarios:
1. People optimized their routes until it the "costs" of increasing that optimization made it less worth seeking out maximal optimization.
2. People found a route that they were satisfied with and didn't give it much more thought.
Some combination of those two explanations strikes me as a good fit for how most people actually approach their work commute. I have a few routes I can take to-and-from the office, but I generally don't give them much thought because it is goddamn 7 in the morning and fuck this traffic and I'm erasing this whole experience from my memory as soon as I get to my destination for the sake of my sanity.
posted by Panjandrum at 9:55 AM on October 6, 2015
Who doesn't optimize their commute and continue to refine it?
If someone is a bad inherent navigator or just doesn't care that much.
Also lots of people learn how to get places socially. So and so took me this route, so this must be the most efficient route because I have respect for so and so.
posted by mayonnaises at 10:12 AM on October 6, 2015
If someone is a bad inherent navigator or just doesn't care that much.
Also lots of people learn how to get places socially. So and so took me this route, so this must be the most efficient route because I have respect for so and so.
posted by mayonnaises at 10:12 AM on October 6, 2015
Is it really a positive that people are spending more money on transportation? How does this affect traffic congestion? pollution?
posted by lownote at 10:20 AM on October 6, 2015
posted by lownote at 10:20 AM on October 6, 2015
Who doesn't optimize their commute and continue to refine it?
This must be down to Citymapper (amazingly good) and similar apps. Android and Apple even send you notifications about this stuff.
posted by colie at 10:23 AM on October 6, 2015
This must be down to Citymapper (amazingly good) and similar apps. Android and Apple even send you notifications about this stuff.
posted by colie at 10:23 AM on October 6, 2015
They should could continue to force people to examine their commute options the way the New York MTA does: randomly take routes out of operation either for certain hours or just eliminate the route entirely. Breaking everyone's commute every 6 months or so guarantees that nobody gets settled.
I can assure you that the London Underground is just as good if not better at this than the New York MTA.
posted by srboisvert at 10:57 AM on October 6, 2015
I can assure you that the London Underground is just as good if not better at this than the New York MTA.
posted by srboisvert at 10:57 AM on October 6, 2015
They don't state their assumptions here, which is problematic, and could easily lead to an incorrect conclusion. They say 5% of people benefited in terms of a lower commute time, but there's no range in terms of either the average cost to everybody of the strike, nor of the average savings to commuters who changed. It obviously makes a difference if everybody lost five minutes or two hours during the strike, and whether it saves you 20 seconds or 20 minutes on your new commute. Presumably they know those values, and used them to draw their conclusion.
What does matter - and different assumptions could easily be made - are:
How much time was spent and saved? They don't actually know this, even though they pretend they do. They know how long people spent from when they entered the public transport system until they left. This morning, the weather was lovely and I was in no rush to get to work, so I walked to the 40th street station instead of the 45th. (I work at 5th, and live at 44th)*. My commute took me more time, but the on-the-train part of my commute was shorter; this is all the researchers would see, and this matters when you are adding up infinitesimals. If lots of people walked when they would have taken the train during the strike, can you count that time up using only public transport on-off information? And what of the time and money spent on the uncertainty around the strike? Time spent planning to avoid disruptions, meetings and business lost or postponed, extra expenditures of time or money on contingencies, opportunities missed entirely to avoid travel. This doesn't seem to be included at all.
How long does the payback period take? If you assume long enough, you could justify almost any disruption as long as someone benefits in perpetuity. The relative duration of the payback period is almost certainly shorter than most would imagine; as some here have mentioned, as soon as a bus or train schedule changes, your efficient route could be made less so. Even if this wasn't the case, the housing and labour markets are ever more in flux; they could have observed someone adjust to a faster commute in February, only to change jobs, to retire, to move house, to change their working hours, or to have their workplace move in March, rendering the benefit of the faster commute moot. One of the challenges with public transportation is it is very time dependent - the best route to get there at 7:30 AM is not necessarily the best route to get there at 6:30 AM, or even for 7:40 AM. I suspect they are not considering the frequency with which work conditions change in the modern workplace; the average time for payback may be less than a year.
What was the value of the time spent and saved? The easiest assumption is that a minute is a minute; if a trip takes a minute longer during the strike, it is made up for by saving a minute after the strike. I suggest that in many cases, a minute is not a minute. In particular, a small fraction of time saved during a regular commute may not be highly valued. If your train ride took 10 seconds longer each day, you might never notice; that's only a sentence or two more in your book or podcast. But that could be added up to almost 30 hours over a twenty year time period. On the other hand, a long delay or a long walk in miserable February weather; missing a meeting because you are running very late or making a lot of preparations to avoid troubles - those could be very unpleasant and costly minutes indeed.
Now, if you make one set assumptions on these, you wind up with an exciting novel paper with a counterintuitive result and broad publication; if you make a slightly different set of assumptions on these, you wind up with an extremely tedious paper showing the obvious that no one will ever be interested in. One of the most fundamental principles of economics is that people will make decisions and respond rationally to incentives; I suspect it applies to these economists as much as it does to anybody else.
*Actual addresses modified for clarity and anonymity.
posted by Homeboy Trouble at 11:44 AM on October 6, 2015 [3 favorites]
What does matter - and different assumptions could easily be made - are:
How much time was spent and saved? They don't actually know this, even though they pretend they do. They know how long people spent from when they entered the public transport system until they left. This morning, the weather was lovely and I was in no rush to get to work, so I walked to the 40th street station instead of the 45th. (I work at 5th, and live at 44th)*. My commute took me more time, but the on-the-train part of my commute was shorter; this is all the researchers would see, and this matters when you are adding up infinitesimals. If lots of people walked when they would have taken the train during the strike, can you count that time up using only public transport on-off information? And what of the time and money spent on the uncertainty around the strike? Time spent planning to avoid disruptions, meetings and business lost or postponed, extra expenditures of time or money on contingencies, opportunities missed entirely to avoid travel. This doesn't seem to be included at all.
How long does the payback period take? If you assume long enough, you could justify almost any disruption as long as someone benefits in perpetuity. The relative duration of the payback period is almost certainly shorter than most would imagine; as some here have mentioned, as soon as a bus or train schedule changes, your efficient route could be made less so. Even if this wasn't the case, the housing and labour markets are ever more in flux; they could have observed someone adjust to a faster commute in February, only to change jobs, to retire, to move house, to change their working hours, or to have their workplace move in March, rendering the benefit of the faster commute moot. One of the challenges with public transportation is it is very time dependent - the best route to get there at 7:30 AM is not necessarily the best route to get there at 6:30 AM, or even for 7:40 AM. I suspect they are not considering the frequency with which work conditions change in the modern workplace; the average time for payback may be less than a year.
What was the value of the time spent and saved? The easiest assumption is that a minute is a minute; if a trip takes a minute longer during the strike, it is made up for by saving a minute after the strike. I suggest that in many cases, a minute is not a minute. In particular, a small fraction of time saved during a regular commute may not be highly valued. If your train ride took 10 seconds longer each day, you might never notice; that's only a sentence or two more in your book or podcast. But that could be added up to almost 30 hours over a twenty year time period. On the other hand, a long delay or a long walk in miserable February weather; missing a meeting because you are running very late or making a lot of preparations to avoid troubles - those could be very unpleasant and costly minutes indeed.
Now, if you make one set assumptions on these, you wind up with an exciting novel paper with a counterintuitive result and broad publication; if you make a slightly different set of assumptions on these, you wind up with an extremely tedious paper showing the obvious that no one will ever be interested in. One of the most fundamental principles of economics is that people will make decisions and respond rationally to incentives; I suspect it applies to these economists as much as it does to anybody else.
*Actual addresses modified for clarity and anonymity.
posted by Homeboy Trouble at 11:44 AM on October 6, 2015 [3 favorites]
Yeah, but if not for the strike, you would not have that delicious plate of beans, homeboy trouble, and neither would we.
posted by notyou at 12:41 PM on October 6, 2015
posted by notyou at 12:41 PM on October 6, 2015
You guys buy this?
Maybe brits are a little set in their ways? Who doesn't optimize their commute and continue to refine it?
I'm gonna take a guess that you've never visited London.
A lot of people who move to London start off using the tube to get everywhere. After you've been there a while you tend to figure out that other options such as busses can be better for various reasons - you just have to know the routes. It's entirely believable that the underground strike could precipitate a bunch of people having to figure this out quicker than they otherwise would have done.
posted by iotic at 4:40 PM on October 6, 2015 [2 favorites]
Maybe brits are a little set in their ways? Who doesn't optimize their commute and continue to refine it?
I'm gonna take a guess that you've never visited London.
A lot of people who move to London start off using the tube to get everywhere. After you've been there a while you tend to figure out that other options such as busses can be better for various reasons - you just have to know the routes. It's entirely believable that the underground strike could precipitate a bunch of people having to figure this out quicker than they otherwise would have done.
posted by iotic at 4:40 PM on October 6, 2015 [2 favorites]
Homeboy Trouble: How much time was spent and saved? They don't actually know this, even though they pretend they do. They know how long people spent from when they entered the public transport system until they left. This morning, the weather was lovely and I was in no rush to get to work, so I walked to the 40th street station instead of the 45th. (I work at 5th, and live at 44th)*. My commute took me more time, but the on-the-train part of my commute was shorter; this is all the researchers would see, and this matters when you are adding up infinitesimals. If lots of people walked when they would have taken the train during the strike, can you count that time up using only public transport on-off information?
Yes, but remember: they didn't just look at who changed their routine during the strike itself - when, it's true, someone may have spent more time walking before entering then exiting the system, thus showing a shorter journey time for the public transport segment of their journey - but also those who continued to use that shorter-on-public-transport route after the strike had finished.
The data was anonymised but individually-trackable, so they'd be able to see the following type of data: prior to the strike, D Jones (not their real name), Oyster card number 056331970128 (not real card number), prior to the strike entered at station A at 07:33 and out at station B at 08:03.
During the strike: tap in at station C, 07:41, tap out at station B, 08:04
After the strike: tap in at station C, 07:41, tap out at station B, 08:04
The fact that each of these individuals kept using their altered route even when the pre-strike route is again available indicates that the time savings persist, even if the data doesn't include walking time, because although the study didn't have access to that data, the individuals being tracked certainly did (I know how long it takes me to walk somewhere), so if walking time outweighed time saved on the train part of the commute, you would very much expect everyone involved to return to their previous route. That 5% didn't is a very strong hint that said 5% found a quicker route to work, even if it involved a slightly longer walk.
posted by Len at 11:10 PM on October 6, 2015
Yes, but remember: they didn't just look at who changed their routine during the strike itself - when, it's true, someone may have spent more time walking before entering then exiting the system, thus showing a shorter journey time for the public transport segment of their journey - but also those who continued to use that shorter-on-public-transport route after the strike had finished.
The data was anonymised but individually-trackable, so they'd be able to see the following type of data: prior to the strike, D Jones (not their real name), Oyster card number 056331970128 (not real card number), prior to the strike entered at station A at 07:33 and out at station B at 08:03.
During the strike: tap in at station C, 07:41, tap out at station B, 08:04
After the strike: tap in at station C, 07:41, tap out at station B, 08:04
The fact that each of these individuals kept using their altered route even when the pre-strike route is again available indicates that the time savings persist, even if the data doesn't include walking time, because although the study didn't have access to that data, the individuals being tracked certainly did (I know how long it takes me to walk somewhere), so if walking time outweighed time saved on the train part of the commute, you would very much expect everyone involved to return to their previous route. That 5% didn't is a very strong hint that said 5% found a quicker route to work, even if it involved a slightly longer walk.
posted by Len at 11:10 PM on October 6, 2015
I live a 10-minute walk from two subway stations on two subway lines, and I walk places. I used to take taxis, until Uber Black became a thing and I could have a stranger's non-taxi car waiting outside my door rather than hailing a taxi two stoplights from the subway station, and now I only take subways at rush hour when nothing with 4 wheels on the surface can even move. I also often take the bus from popular bar districts where getting a taxi is difficult and Uber people won't even pick up I mean what. I have to ride the bus 2-3 stops, and then I can get Uber. Maybe to a subway station if it's rush hour.
I also have a folding bike with a flat tire. I'll have to put it in a taxi to bring it to the store to fix it, won't I?
I work from home. Five minutes away from a place that sells watercress salad and hand-made sausages.
Try to optimize me even I bet you can't I dare you. I DARE YOU.
posted by saysthis at 5:19 AM on October 7, 2015
I also have a folding bike with a flat tire. I'll have to put it in a taxi to bring it to the store to fix it, won't I?
I work from home. Five minutes away from a place that sells watercress salad and hand-made sausages.
Try to optimize me even I bet you can't I dare you. I DARE YOU.
posted by saysthis at 5:19 AM on October 7, 2015
It's cool. Labour self-determination is good for society. I don't need petty analysis of transit route patterns to continue believing that.
But it does tickle the nerdbrain nicely...
posted by rum-soaked space hobo at 3:44 PM on October 7, 2015
But it does tickle the nerdbrain nicely...
posted by rum-soaked space hobo at 3:44 PM on October 7, 2015
« Older But where's the Coffee Shop AU? | BM 2.0 Newer »
This thread has been archived and is closed to new comments
posted by miyabo at 9:00 AM on October 6, 2015 [2 favorites]