Tracking job and housing dynamics with smartcard data | Proceedings of the National Academy of Sciences (open access)

Recently published:

  • Jie Huang, David Levinson, Jiaoe Wang, Jiangping Zhou, and Zi-jia Wang (2018) Tracking job and housing dynamics with smartcard data  (Open Access) 




This paper uses transit smartcards from travelers in Beijing retained over a 7-y period to track boarding and alighting stations, which are associated with home and work location. This allows us to track who moves and who remains at their homes and workplaces. Therefore, this paper provides a longitudinal study of job and housing dynamics with group conceptualization and characterization. This paper identifies four mobility groups and then infers their socioeconomic profiles. How these groups trade off housing expenditure and travel time budget is examined.



Residential locations, the jobs–housing relationship, and commuting patterns are key elements to understand urban spatial structure and how city dwellers live. Their successive interaction is important for various fields including urban planning, transport, intraurban migration studies, and social science. However, understanding of the long-term trajectories of workplace and home location, and the resulting commuting patterns, is still limited due to lack of year-to-year data tracking individual behavior. With a 7-y transit smartcard dataset, this paper traces individual trajectories of residences and workplaces. Based on in-metro travel times before and after job and/or home moves, we find that 45 min is an inflection point where the behavioral preference changes. Commuters whose travel time exceeds the point prefer to shorten commutes via moves, while others with shorter commutes tend to increase travel time for better jobs and/or residences. Moreover, we capture four mobility groups: home mover, job hopper, job-and-residence switcher, and stayer. This paper studies how these groups trade off travel time and housing expenditure with their job and housing patterns. Stayers with high job and housing stability tend to be home (apartment unit) owners subject to middle- to high-income groups. Home movers work at places similar to stayers, while they may upgrade from tenancy to ownership. Switchers increase commute time as well as housing expenditure via job and home moves, as they pay for better residences and work farther from home. Job hoppers mainly reside in the suburbs, suffer from long commutes, change jobs frequently, and are likely to be low-income migrants.

How more development can lead to less travel: Examples

Balancing housing and jobs, so that they are located near each other, logically reduces travel compared to a situation where those same jobs are far apart. This has long been understood in the transport planning community (see e.g. Cervero 1989, or my 1998 paper), but is not well grasped among the general public.

However, moving a fixed number of things around is not how cities actually grow. Telling place A you taking away their employment is controversial. More generally new things are added.

Development in Mascot. Photo by author.
Development in Mascot. Photo by author.

It is commonly asserted that more development adds to congestion. And often this is true. But not always, it depends on the type of development. More housing in a housing-rich and job-poor area will result in more total travel. More employment in a job-rich, housing poor area will do similarly. More housing in a job-rich area, and more jobs in a housing-rich area can actually reduce travel.

For our baseline case, imagine a city with two precincts separated by 2 km.

Precinct A: 1000 Jobs, 0 Resident Workers

Precinct B: 0 Jobs, 1000 Resident Workers.

The one-way (morning commute) trip table looks like:

Jobs 1000 0
Workers A B
0 A 0 0
1000 B 1000 0

Total daily travel to work is 2000 person km per day. (Everyone commutes from B to A). Travel on Link BA is 1000 at 2 km per trip, or 2000 person km traveled. (This just analyzes one-way trips. Round trip commutes would double this.)

Case 1. 

There is a proposal to intensify development in Precincts A and B, so each is more locally balanced.

Precinct A: 1000 Jobs, 500 Resident Workers

Precinct B: 500 Jobs, 1000 Resident Workers.

The new one-way (morning commute) trip table looks like (rounded):

Jobs 1000 500
Workers A B
500 A 498 2
1000 B 503 497
  • assuming 0.5 km intrazonal travel distance, using a doubly-constrained gravity model with a d_{ij}(-2) impedance function.

The Daily Travel on links:

AB = 2 @ 2 km

BA = 503 @ 2 km

within A = 498 @ 0.5 km (walking)

within B = 497 @ 0.5 km

TOTAL = 1507 pkt.

This is considerably less than the baseline case as many more travelers can reach their destinations locally. While there is still some commuting, it is far less than before.

Case 2.

There is a proposal to build a locally-balanced Precinct C halfway between Precincts A and B.

Precinct C has 500 Jobs and 500 Workers

The new one-way (morning commute) trip table looks like:

Jobs 1000 0 500
Workers A B C
0 A 0 0 0
1000 B 666.666667 0 333.333333
500 C 333.333333 0 166.666667
  • assuming 0.5 km intrazonal travel distance, using a doubly-constrained gravity model with a d_{ij}(-2) impedance function.

The Daily Travel on links:

BC = BA + BC = 1000 @ 1 km

CA = BA + CA = 1000 @ 1 km

within C = 166 trips @ 0.5 km

TOTAL = 2083 pkt.

In this example, the total person kilometers traveled (pkt) on the links connecting inter-city precincts is essentially identical to the base case, despite adding 500 residents and 500 workers halfway between each. There are an additional 167 pkt daily on the intrazonal market (within C), which is likely walking.

The total one-way commute travel per person however drops, from 2 km/person per day to about 1.38 km/person per day. The average trip length is reduced. The experienced travel is thus about one-third lower.

Case 3

Building on Case 1, completely balancing A and B (so each has 1000 jobs and 1000 workers) reduces one-way commutes further (to 1176 pkt)

The new one-way (morning commute) trip table looks like (rounded):

Jobs 1000 1000
Workers A B
1000 A 941 59
1000 B 59 941
  • assuming 0.5 km intrazonal travel distance, using a doubly-constrained gravity model with a d_{ij}(-2) impedance function.

So, it should be clear from this example that adding development can actually reduce total travel, if it is the right kind of development in the right places.

The End of Traffic and the Future of Access (Free)

The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape. By David M. Levinson and Kevin J. Krizek.
The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape. By David M. Levinson and Kevin J. Krizek.

We are pleased to announced that you can now download a PDF of The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape from the University of Sydney eScholarship Repository (Free).

Title: The End of Traffic and the Future of Access: A Roadmap to the New Transport Landscape
Authors: Levinson, David
Krizek, Kevin J.
Keywords: transport
automated vehicles
sharing economy
Issue Date: Oct-2017
Publisher: Network Design Lab
Citation: Levinson, D. M., & Krizek, K. J. (2015). The End of Traffic & the Future of Transport. Network Design Lab.
Abstract: In most industrialized countries, car travel per person has peaked and the automobile regime is showing considering signs of instability. As cities across the globe venture to find the best ways to allow people to get around amidst technological and other changes, many forces are taking hold — all of which suggest a new transport landscape. Our roadmap describes why this landscape is taking shape and prescribes policies informed by contextual awareness, clear thinking, and flexibility.

If you want other versions, please go here.

The Alexandria – Moore Park Disconnector | WalkSydney

I wrote The Alexandria – Moore Park Disconnector for WalkSydney about a road widening project in my neighbourhood.

The help it needs is not making it wider so cars can speed through from not here to not there, but making it narrower, so people  on foot can cross from shops on one side to shops on the other. In other words, it should be moved down rather than up the hierarchy of roads.  People should be able to cross these streets freely and without fear. McEvoy should be a destination, not a detour

Read the whole thing, and if you are a New South Welshman, feel free to get in touch to help make Sydney a better place.


We are pleased to help launch … the website of WalkSydney, a new group supporting a more pedestrian Sydney. Visit for more details and the latest posts.

Transport Accessibility Manual Working Group

I am pleased to announce that the Transportation Research Board Annual Meeting will host the first in-person meeting of the Transport Accessibility Manual Working Group.


Transport Accessibility Manual Working Group

Monday 6:00 PM- 7:30 PM
Marriott Marquis, Mint (M4)

David Levinson, University of Sydney, presiding

Sponsored by:

Standing Committee on Bus Transit Systems (AP050)

One of the key problems is what to value when investing in transport or regulating land development. Accessibility — the ease of reaching valued destinations — connects transport and land use, considering both how easy it is to move and where things are located. While many planners know how to measure this, many don’t, and all could benefit from standardizing application to best practice. To that end, a working group would develop such a standard, which would clarify topics like how to measure, how to compute, how to present, and what to consider in terms of accessibility.



If you are interested in participating, please email me.

Sydney’s Refugee Crisis | WalkSydney

I wrote a thing for WalkSydney: Sydney’s Refugee Crisis. Read the whole post.

Refugee children, trapped on a small island, trying to get to a better place, are challenged by a convoy of heavily armoured, multi-ton metallic objects moving within a mere 1 meter past their developing bodies at 50 km/h. What could go wrong?


Overall, refuge islands and free lefts are terrible traffic engineering and worse urban design. Instead, narrow the intersection and design it so pedestrians are controlled by only one light when crossing the street. This is especially pernicious at this location, with schoolchildren saturating the island.

At least 20 people, most of them children, queued up like bowling pins, on a refugee island at City Road and Cleveland Street, Sydney. Other groups are held back at the Seymour Centre, or have been already ferried across City Road to Victoria Park. This is typical, but not good, urban design.
At least 20 people, most of them children, queued up on a refugee island at City Road and Cleveland Street, Sydney. Other groups are held back at the Seymour Centre, or have been already ferried across City Road to Victoria Park. This is typical, but not good, urban design.

The Transportist: November 2018

Welcome to the November 2018 issue of The Transportist, especially to our new readers. As always you can follow along at the  blog or on Twitter







Automated, Autonomous, Driverless, and Self-Driving Vehicles, and Semi-Autonomous Systems 

Electric Vehicles [and Renewable Energy]

Human-Driven Vehicles, Signs, Signals, Sensors, and Markings, and Roads


Shared Vehicles/Ride-sharing/Ride-hailing/Taxis/Car Sharing


Human-Powered Vehicles/Bikes/Pedestrians/Scooters/eBikes/Last-Mile/First-Mile/etc

Land Use/Architecture

Kerbs and Sidewalks

Retail, Freight, Waste, and Logistics

Technology History

Travel Behaviour

Intercity Trains

Aviation and Space




Research & Data

Papers by Us


by Others



Other People’s Newsletters

Read the Transportist, but don’t just read the Transportist, also read:


On Academic Compliance Bullsh*t

Harry Frankfurt wrote a book “On Bullshit“, which Wikipedia summarizes as saying “bullshit is speech intended to persuade without regard for truth.” I think the problem is deeper than that. There is work generated for the sake of saying that work was done.

Peer Review

Consider peer review. I recently received a review from a paper I co-authored in a good journal. The reviews were positive except Reviewer #2 said the word X was not the right word. X is of course exactly the right word, but in order to get accepted we had to change the paper to make Reviewer #2 happy. [I refuse to accept the charitable view that these were Reviewer #2’s genuine beliefs, it is truly nonsense.]

We complied. We wasted our time to increase the utility of anonymous Reviewer #2 in order to satisfy the editor. Reviewer #2’s ego is boosted, by having enforced compliance, and therefore increasing his relative status at the expense of ours, but since he is anonymous, only he knows. Reviewer #2 could have just said “Accept”, but that would be too easy, he felt he had to say something to prove he reviewed the paper. (R2 could be female, but he feels male.)

Now Reviewer #2 is not operating in a vacuum. Undoubtedly some unreasonable reviewer of one of his papers made him go through what he felt were ridiculous contortions, and this rolls downhill. To salvage ego, the abused child becomes an abuser, creating a new generation of abused spouses and children.

What we have here is a cycle of peer review violence, where as more and more research is produced due to increased productivity of academics (in part due to the rise of information technologies, but mostly the publish or perish culture driven by university ranking systems driven by the desire to attract international students driven by revenue), more review requests are generated, more reviewers get more annoyed at the requests, and more hoops laid out before us.

This Reviewer #2 was actually not so bad. Many others are unhappy if you don’t regurgitate all scientific knowledge up until the present day, and lay out all prospective policy outcomes going forward. This attitude has led to an explosion of paper lengths.

Reviewers would be much more polite were reviews not anonymous. This raises other problems, that junior people would be afraid to confront senior people on cases which were not bullshit. Unless everything is open, an open (non-blind) review policy from a single journal cannot  extract fair reviews. Retaliation, on say grant reviews, or promotion reviews, which remain anonymous, is a risk.

Source: Are economics papers too long?

Without any peer review, Gresham’s Law of Journal Articles: Bad knowledge drives out good, would surely apply. Peer review of some form is a Good Housekeeping Seal of Approval for scientific articles. But that said, what is really required here? If you, a journal, trust me enough to review (or edit) other people’s work, why do you not trust me enough to publish my own? There are a few arguments in favor of review:

  • First, we have Linus’s Law: “with many eyes all bugs are shallow”, and so some editorial review will improve quality and find problems. Good authors want good editors.
  • Second, the anticipation of peer review improves quality as you know a paper will have to get through review to be published

However in contrast, if we have a peer review system where nothing passes the first round (regardless of how good), but many papers go into the revise and resubmit limbo, authors will, in fact, submit lower quality work, and wait for the reviewers to make their recommendations, and spend their scarce time trying to satisfy those reviewers instead of themselves.  In short, we have constructed a system where peer review lowers the initial quality of submission. We have become so afraid of publishing false positives (a wrong paper), we create many false negatives (decline competent papers). History can judge false positives retrospectively just fine, we don’t actually need to spend so many resources to do this prospectively.

I have talked previously about how peer review also costs society knowledge, as the inevitable delay in the Revise and Resubmit round, and the cost of going back to closed projects.  Instead of rewarding academics by which journals they published in, reward them for how important their work is. This is either known, because history rewards them with citations, or arguable that the future will recognize them because colleagues believe in them now.

Instead of over-reliance on peer-review, we should view it as a filter to ensure wrong or poorly written papers are not published, not a filter to ensure only perfect papers are published. We should have a system that rewards the creation of small (or large) academic building blocks, and lets scientists and engineers and even economists file their work respectably as they develop it in the length appropriate, and not feel the need to expand their work to develop a whole new theory of civilization with every research output.

History can be the evaluator – it is attention which is now scarce, not the number of pages in a journal. Compliance with systems built for another age needs to be tossed with those systems.


Seeking letters for promotion cases aims to ensure that an outsider (someone not at your university) says you do good work, because for some reason, the university cannot trust people at its own university to make such a judgment. My promotion cases required 10 or 11 letters from other academics at other universities saying that my work was good enough to warrant promotion, and I have written numerous anonymous letters. I have not retaliated (nor had the opportunity to ‘retaliate’ against a youngster I was offended by, or their senior allies) by trying to undermine a promotion case, but I certainly can see how some senior people might if they were offended by a junior faculty somewhere, or by those junior’s senior colleagues, like a peer review.

I understand that such letters help assure that promotion is warranted, but imagine Apple computer asking Microsoft, Google, and Facebook to write letters in support of promotion of their own software engineers. That’s absurd.  The evidence of my research is in my publications, and other people’s citations of those publications, not in whether someone else says my work is good. My colleagues should be able to judge that. The evidence of my teaching is in whether my students learned (and retained) anything, not in end-of-semester surveys.

But if I as a junior faculty know that I have to get 10 senior people to write letters for me, I will spend effort to curry favour by doing things like reviewing papers when assigned by editors, and serve on committees, and so on. In short, I will comply in advance so that the favour will be returned. Letter writing enforces compliance on the part of junior faculty structurally.

Let universities take on the burden themselves of deciding whom they should promote, rather than offloading this to the community. If they don’t feel comfortable assessing their own staff, maybe that’s a field they shouldn’t be in.


Sometimes compliance-enforcement takes an even more ridiculous turn. I recently had a conference paper at an Australasian conference accepted on its substance but declined because of some mysterious MS Word formatting problem that I refused to spend even more time to rectify after 2 previous revision attempts. Despite using the organisers templates, they still decided the paper somehow didn’t meet the correct format, and so was rejected. Obviously it’s their loss, they’ll miss me, and our research,  (and my student who would have also presented something else), and the revenue we would have paid to attend the conference. If the papers were to be published in a book, I might understand why this matters, but that in fact was not the case, it was simply for electronic distribution, and the aesthetic judgment of the organiser, which is lacking (obviously, as it was a pretty ugly MS Word template to begin with).


Now I understand Tyler Cowan’s quote (can’t find the original, but essentially)

 “The most important thing I learned in my PhD was to get the margins on the pages right.”

Back in the day, an older woman in the registrar’s office would go through your thesis or dissertation with a ruler and measure to make sure the margins on each page were just so. And if not, the dissertation would be turned back, and you got to reformat it. This was the University’s final lesson in compliance.

But really, why does marginal perfectionism matter? We did it because the system required it. Fortunately, this particular requirement has disappeared, but why did the system require it? One imagines so that reproductions of the dissertation on a smaller sheet of paper would not lose important information. There may have once been a good reason, but margin size enforcement was promulgated as a rule that lasted long past the original need.

Establishing Degrees

At a major university with which I have an affiliation, I am working on establishing a new degree program. This is a relatively cost-free enterprise the university, the units of study are almost entirely already offered. However to get the program established, we have to have an Expression of Interest, vetted by three faculties involved, including 2 committees in my faculty, as well as two committees at the university level. Then we have a proposal, whose form is 57 pages. And then we need to go through all the same committees. I am told the 57 page form is designed to dissuade people who aren’t serious. But for those who are, that and all the meeting for something so technically simple to implement is pointless.

One of the faculty committees has about 50 members, all of whose job, apparently is to ratify what the other committees said and supervise one or two full-time staff members.

Let a thousand degrees flourish, and if they don’t succeed, they can be cancelled.


Just as universities accredit students (who undoubtedly think exams and homeworks and projects are a nuisance), degree programs often go through accreditation themselves to show that the curriculum they require students engage in comports with what the industry associations who control ABET think is important (or was important, as this is an exceptionally conservative process designed to stifle innovation.) The last time I went through this (fortunately I did not have to lead it in my Department), each required course produced a notebook with sample poor, average, and good work from students for each assignment, as well as printouts of the assignments and other miscellany. It is pretty clear the review panel did not actually review the contents of each notebook. They may have sampled them.  The  wall of notebooks was there to demonstrate compliance. Each assignment was cross-referenced against objectives and qualities students were supposed to accomplish by successfully completing that assignment. While this sounds good in principle, it is basically a database exercise, labelling things as satisfying objectives rather than changing things to meet objectives.

Let universities produce students whose value is they graduated from a university that taught what it thought important, and if that aligns with market demand, all the better. It is not as if students don’t also have to take and pass exams to be a Professional this or that, and an education that helped would be appropriately recognized, or universities don’t have well-established and largely self-fulfilling reputations.


Academics do nothing if not evaluate each other’s work. The amount of time writing letters of recommendation, evaluations of promotion cases, reviews of proposals and each other’s programs, and conducting peer reviews  of articles is surprisingly, and in my view unnecessarily, high. It is academia generating work for academics who ought to be in the primary business of creating and transmitting knowledge, not evaluating knowledge creation and transmission. It is, in economic terms, a deadweight loss. If all this evaluation improved the quality of knowledge production or transmission sufficiently, it might not be, but there is no evidence I see such is the case. We adopt the forms because those before us adopted the forms.