Hacker News new | past | comments | ask | show | jobs | submit login
Walkability a key factor determining upward mobility of a city's residents (bipartisanalliance.com)
244 points by Fricken on Oct 5, 2019 | hide | past | favorite | 69 comments

Paper: https://www.researchgate.net/publication/329729827_The_Socio...

So they buy/beg data for 382 "community zones" from walkscore.com, do linear regression, and find a correlation. Then they get funding to do an individual-level survey and there's no correlation. And it doesn't appear in South Korea either.

Their explanation is some statistical magic to pull out a "sense of belonging" as a mediating factor and a statement in the conclusion that "It may be that walkability is an emergent property most clearly visible at the level of aggregate, not at the level of each individual."

No discussion of data accuracy besides noting that even perceived walkability and actual amount of walking are barely correlated in their surveys (i.e., their data sucks).

Interesting paper idea, but the implementation seems like more statistical garbage.

To quote the study, the walkscore.com data is "the earning records from all American citizens born between 1980 and 1982 whose parents filed taxes." Your use of the number 382 seems to be rhetorical implying that the sample isn't large enough. Moreover, your use of "buy/beg" verges on a predicative sort of ad hominem. Your criticism are totally valid, but you missed the most egregious flaw: "[O]ur analyses treat walkability and access to public transportation as interchangeable."

“whose parents filed taxes.“

As an outsider to the American tax system: is this a strong selection that could impact results? Here in Germany for example filing is optional for a lot of people (not just legally, but also in terms of expected returns) which would have quite unpredictable results on a study like this.

whose parents filed taxes

Excludes recent immigrants or orphans then, both of which groups could be particularly acute case-studies of upward mobility.

Big picture though, walkability and public transport definitely imply at least the potential for interaction and empathy, and cars the opposite.

The Highway and Automobile culture are symbols of totalitarian cultures which deny people more sustainable and equitable alternatives for mobility and transport. - Vandana Shiva, February 19, 2004.

It is 100% mandatory to file taxes if you make over $12k and over $13.5k if you’re over 65 and file single. Doesn’t matter if it’s from a job, investment, or unemployment.

I can imagine Americans living abroad will be shaking their fist at the parent poster. They have to pay taxes in the USA even when working abroad, never mind in the USA!

Sigh, this always comes up and I’ve never met a single person who actually had to pay taxes. My brother in law was out of the country for a decade and a few other friends lived in Europe and never paid.

Has anyone on HN paid US taxes living abroad?

You do still have to file a tax return, which can be a bit complicated if you have foreign income. But if you work in Europe, you're right that you likely won't have to pay any actual US taxes. You can take the European tax paid as a foreign tax credit against the US taxes, and that almost always zeroes it out, since European countries mostly have higher tax rates. That was the case for me, anyway.

You will have to pay some tax if you work in a country whose tax rates (for your bracket) are lower than US federal tax rates. That can be the case for Americans working in places like Hong Kong or the Bahamas.

This is often also the case in Switzerland.

California’s residency requirements are quite strict, and they don’t offer foreign tax credits, so if you’re planning to go back after a few years you’re effectively being double-taxed on the state income.

I worked with a guy in the UK who constantly lamented this.

It is required in the states but could potentially limit results to include less low income/unemployed individuals. Almost any job will withhold income tax and send it off to the government without you doing anything.

The walkscore data is separate from the earnings records; it's calculated almost on-demand from the Google Maps API. The earning records are from https://opportunityinsights.org/paper/land-of-opportunity/, they didn't actually analyze the records themselves. And yes, 382 isn't particularly fine for geography of the US: https://opportunityinsights.org/wp-content/uploads/2018/04/m.... For comparison, there are 3142 counties and they could have used that data instead.

Interesting. I know a lot of the statistics have been discussed elsewhere here, so I'll add two weak speculations -

Cars are pretty expensive to own and operate. Mr. Money Moustache is constantly railing against them - https://www.mrmoneymustache.com/2013/04/22/curing-your-clown.... Agree or disagree with his approach to things, his blog is very much about high social mobility.

Secondly, walking is a pretty good way to force yourself to get exercise. I'm fortunate enough to work for a big tech company with a shuttle; I recently realized that taking the shuttle forced me to walk more than 500 miles a year just between the stop and my apartment. That's huge when I'm pretty unreliable about going to the gym. The shuttle also strongly encourages me to stay on a reasonable schedule - in before the last morning shuttle, out before the last evening one. I have to imagine that over time, this is also a pretty huge health benefit.

Edit: typos

I was thinking recently about whether it would be useful for people to start thinking of things in terms of $/mi instead of mpg. Would make it much more explicit that it costs money to go places.

Also as a bit of a side tangent, one of my big pet-peeves is people who state their mpg in best case scenario and not day to day. "Oh my SUV gets 45mpg cruising down the interstate." Yea sure it does. but overall you're still getting low 20s if you measure fill up to fill up.

That depends on the current price of petrol which changes often.

And the price of travel changes as the price of gas changes.

I was taught to check my mileage every time I fill up the tank. It would be just as easy to divide `cost / trip meter` vs `trip meter / gallons`, and would arguably give a more interesting number.

Mostly just an idea I had that I thought may be worth trying.

Walkability goes hand-in-hand with density. Density tends to occur close-in as a consequence of land values being higher.

Therefore, there is a natural association between high property values and walkability.

I hope the paper takes this into account. It could just be that people with more money live in more desirable locations, which end up being more walkable too.

Dense cities being expensive is relatively recent in the United States. It follows decades of restrictive zoning and changes at the federal level to make housing more of an investment. Also, don’t forget that for most of history, you could only travel on foot, by horse (or similar), or by boat. “Walkability” is a retronym; for the longest time it was the only way most people got around.

Edit: Downvotes? Stay classy, HN.

I agree some of the reasons for density are artificial. Not all of them, because the more desirable an area is, the more infill makes sense. (You see vacant lots in rundown areas, not thriving ones.)

But yes, that's a real issue, and it's a disastrously bad policy if you care about the environment (because density reduces transportation energy use).

However, I think the artificial restriction only magnifies the effect here. Lots of people would like to live in a dense, walkable area. If those policies weren't in effect, most people who wanted to could. But since they are, density is for those who can afford it.

Dense cities being expensive is also mostly a relatively recent phenomenon in the US because until about 20 years ago there was a next outflow of population in most cities.

Supply is certainly part of the problem for expensive housing in some cities. But it's also driven by a significant increase in demand over a fairly short period. (And IMO it remains to be seen whether trends like remote work, autonomous vehicles, etc. end up reversing the inflow into "elite" cities.)

Right, but doesn't this explain what they see just fine? The walkable parts of cities are the gentrifying cores, which are now expensive, but 30 years ago were in deep decline. Were poor, now rich, call that upward mobility to sell the paper?

Other than that they're largely different groups of people. It's mostly young professionals displacing blue collar workers and racial minorities.

With some exceptions like Manhattan, graduates of 20 to 30 years ago didn't live in city cores or often even commute to them. In my grad school class--the New Yorkers aside--almost no one I knew moved into a city. They were all out in the suburbs, which is where most of the companies they worked for were located as well.

Yea, I know. And since they track people by tax returns it can't literally be the rich newcomers who are the today datapoint. But I'm still pretty dubious about the causality they claim.

News Guidelines:

Please don't comment about the voting on comments. It never does any good, and it makes boring reading.


Seems spurious. Could be talented people move to cities with high earning potential—which happen to be walkable. Do the authors control for that?

Agreed. The most desirable and expensive neighborhoods tend to be the most walkable ones. It would make sense that the people who live in those neighborhoods tend to be more financially successful.

It's as if a study came out that showed that, according to their exhaustive statistical analysis, people that drove Mercedes tended to be wealthier than people who drove Chevys. Then they concluded that the key to upward mobility must be to get a Mercedes.

It's supposed to be mobility, and I think tracking individuals through their tax returns. (This magic data which only Chetty & co have access to.) So I don't think the Mercedes example works.

But I agree that it's very unlikely to be that being able to walk places makes you succeed. Could even be something as simple as enough people owning a few flats in their neighbourhood, which suddenly became super-valuable when lots of google employees showed up. That would look a lot like income mobility.

Yeah, something about this study just feels like it will ultimately end up as an example in an article covering the replication crisis or something.

It's longitudinal tax data.

>We 1st identified the relationship between walkability and upward mobility using tax data from approximately 10 million Americans born between 1980 and 1982.

The data structure doesn’t help with causal inference.

Reading the abstract, it looks like the authors are trying to make a causal argument: people are more likely to improve their economic standing because they moved to a walkable city.

OP pointed out that this could be due to self-selection of people with the potential of upward mobility to walkable cities. This would lead to spurious correlation between economic mobility and walkability. That is, a third variable (a potential candidate, awareness of climate change) could explain both upward mobility (usually, climate change awareness correlates with education level) and economic mobility (education correlates with higher wages).

Without reading the study, this would be my first assumption.

This sounds at best correlation not causation (and the correlation is pretty weak too). I didn't see any causal effect.

My instinct tells me that walkability is a consequence of the actual cause, which is probably higher property values in denser cities, meaning better schools, or something to that effect.

I was living for some time in a city that offered a lot of options for pedestrians but certain things were not available or only with difficulties - e.g. buying furniture - unless you had a car. So living there without a car was fun at first but then it became very draining, especially when having to move from one apartment to the next one.

The logical explanation is probably that with a car one has not only more options but also options are available at lower cost. E.g. one can move to a cheaper area, with cheaper supermarkets but that is only reasonably accessible by car. The irony for me is that cars became that popular in the first place, given that they are a quite unreliable and expensive toy.

Walkability is a proxy for whether a city has a knowledge economy (closely packed buildings) or an industrial economy (large warehouses and factories, less walkable). The knowledge economy cities have far outperformed the industrial cities in economic growth over the last four decades.

By and large, most "downtowns" are primarily knowledge economies whether or not there is a lot of industrial economy centered around a city. If for example you look at the jobs in central Houston or Dallas I think you'll find they are mostly office jobs but those, like many other US cities, are not very walkable except in very limited areas.

Yet it seems common in American cities that the best schools for children are in the suburbs, where walkability is graded lower.

Public schools seem to be regaining credibility as the upper-middle class flood back into cities, but have they caught up? Unsure.

It's less a question of "suburban vs urban" than "wealth vs poverty."

Big cities have wealthy neighborhoods and poor neighborhoods, good schools and bad schools. And likewise, there are wealthy suburbs and poor suburbs.

A school that serves a mix of wealthy and poor suburbs probably has a more-average quality than the low end of a highly segregated city, but you can bet that a school in a tony part of NYC is going to be a hell of a lot better still.

All of this is due to San Antonio vs Rodriguez, which IMHO is the worst Supreme Court decision since Dred Scott.


I was wondering the same thing. How can they isolate walkability rather than cultural factors or factors related to the public school system of a city?

Correlation not causation..nothing to see here.

Lots to see when we find correlation, even if we aren't certain of the causation.

First and foremost, we should attempt to replicate it and see if the correlation holds up independently.

Second, we should attempt to determine if their is a causal relationship. The arrow of causality going in either direction is of great interest to urban planning.

Or third, if there is no causal relationship between them, we should attempt to determine whether there is a third correlating factor, or better still, another factor that has a causal relationship with these two.

This discovery is highly interesting, and points in the direction of useful further research to perform. I would definitely not say that there is nothing to see here.

Authors often do some research, and then present some conjectures. Their conjectures may not be supported by the research, but they point in the direction of further research to confirm or refute the conjectures.

Research like this may not be sufficient, but it may be useful.

There's another possibility: The two variables are completely unrelated but we only get to observe them when some function of both of the variables is true. The classic example is two independent coin tosses that trigger a bell when both coins come up the same -- and we only think to look at the coins when the bell sounds.

Thus my first suggestion that we get independent reproduction of the results :-)

One of the most common such "observability" functions is that humans tend to pay attention to things they find interesting.

Ancedotally, here in Detroit the walkable neighborhoods are the ones gentrifying the fastest and in the process becoming more walkable (more shops, grocers, restaurants)

Average commute distance is 20 miles or so. At $0.25 per mile in depreciation, maintenance, gas, etc... you are spending about $10 per trip. With 260 work days in the year that costs the average American about $3,000 extra at a minimum. If you make less than $30,000 a year that is 10% of your budget gone before even rent is calculated.

In states with higher road taxes and fuel prices, and if you drive a less economical vehicle, that figure can get as high as $10,000 pretty quickly. I'm always surprised at poverty level families driving vehicles that cost them 25-40% of their yearly income. Often they have to use Uber and Lyft to help cover the car's cost and end up depreciating it even faster.

Imagine, being a slave to a freaking car.

I live in a very walkable area. Partly because it’s walkable, it’s also desirable and thus expensive, and tends to attract people who are either wealthy or at the start of their upward mobility - young people from families with means who are fresh out of college and reporting $50k in annual income now, but will likely be reporting $150k in annual income in the next decade. This is the story of most larger northern cities, and it’s a pretty simple explanation for the phenomenon explored here.

In general, the fluidity of movement is a key to a healthy economy.

For example, post 2007/2008 one of the things that slowed the economic recovery was the fact that too many people owned homes. When you own your home you can't just pick up and move (to where the jobs are/growth is). Eventually, some people abandoned their homes. Necessary. Unfortunately, those holes left a mark behind them as well.

There is no opportunity without fluid movement.

Isn’t a city walkable by definition? I am comparing it to a suburb obviously. It feels like they have to define what walkability is first. If there is any correlation it will not be on the overall score but on some aspects that make up the score.

>Isn’t a city walkable by definition?

Not really. Even if you can walk to a restaurant from a company or your apartment/condo/house, if you still have to routinely take cars and/or public transit to do most things, that city isn't very walkable.

A lot of cities have specific areas that are fairly walkable but those areas may be pretty small and/or not particularly mixed use--so you still need to mostly drive.

Here's Redfin's classification scheme: https://www.redfin.com/how-walk-score-works

Correlation is not causation; and public transit is not walking. This is a rat's nest of extreme reaches.

I really like being able to walk places, I think it probably is helping me fiscally, but this is not even a survey.

Ugh oh: correlation is not causation - hence you cannot state it's a "determinant".

They are probably both caused by - say - "criminality" which makes a place more or less walkable.

There is a factor not being considered in walkability studies.

How far/convenient are pharmacies/cheap grocery places.

Those are the most important things

Availability of nearby services such as those is a pretty major component in the walkability scores I've seen like https://en.wikipedia.org/wiki/Walk_Score . I agree they are very important, but I think they are considered. Where do you see them missing?

In real life.

This reads as if there could be an alternative title: “Multi-story city buildings should include stairs.”

They all do for fire safety.

This is not a causal analysis. However, the authors do state this as a limitation in the paper.

Blog post is a copy of the abstract of a $12 article. https://psycnet.apa.org/doiLanding?doi=10.1037%2Famp0000422

Nothing to see here.

I wonder how remote will change this.

London is extremely walkable and doesn’t seem to have that great of a social mobility for its residents.

Do you have better support for this statement than what the authors of cited paper have?

You did, after all, dismiss a whole paper with a glib remark.

One example doesn’t strike me as a glib dismissal, particularly since this study was done in the United States.

They don't give an example, they express their impression rather than providing contrary data.

Do you expect an entire research paper in a single HN comment?

I don't really care either way. I was arguing with your characterization of their comment, not demanding that they retract it.

No, but I would expect HN participants to not dismiss a researched paper with a glib "feeling" that's utterly unfounded.

they didn't give an example but simply made an assertion, one that by the way is likely wrong, given that the East of England is generally upheld as an example of good mobility(https://www.theguardian.com/higher-education-network/2018/fe...), compared to the rest of the country.

Ah, I see. Then I shall counter: London has pretty good upward mobility.

Citation needed. Feels like a deathtrap sometimes.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact