ScPo M1 Research Project Presentation

ScPo M1 Spatial/Trade/Urban Group

Florian Oswald
florian.oswald@sciencespo.fr

SciencesPo Paris

3 March, 2024

Outline

  • Who locates where?
  • What are the global implications of those choices?
  • Ongoing research projects
  • RA-needs

Global Implications: Reims 1950 vs 2016

Ongoing Research



Who Locates Where

  • Classical Urban question
  • Want to apply that in several domains:

1. Parents

Where do parents have children?

2. Family Networks

How does the family network affect location choices?

3. Bus Operators

How do firms locate in space to avoid competition?

1. Fertility, Housing Costs and City Growth

Fertility, Housing Costs and City Growth

  1. There are relatively few children in the center of cities.
  2. Across cities, smaller cities have more children.
  3. In general, there is negative density dependence of fertility across French Urban Areas.

Density Dependent Fertility

Housing Consumption

Why Do We Care?

  • If there is a causal link from house prices to fertility, then…
  • … zoning 🗺
  • … commuting 🚋
  • … labor market 👩🏾‍🎓 👩🏻‍💻 👷🏽‍♂️

are all relevant for 👨‍👩‍👧‍👧 👩‍👩‍👦‍👦 👨‍👨‍👧‍👦

Interesting Tradeoff for a planner: where to incentivize fertility?

  • Urban Econs emphasize agglomeration effects
  • Higher density 👉 fewer children
  • Where to locate parents? (And where other age groups? 👴🏽 👵🏻)

Problem: Joint Decision

Econometrically, that’s a mess: People choose location and fertility at the same time. And if not, highly interdependent.

  • Either need an exogenous shock to location or fertility:

  • Forced location (choices)?

  • Policies randomly affecting fertility?

Or a credible model.

Check it out!

Fertility Model Simulations

Family Networks

Family Networks

  • A lot of work (urban and labor) has considered why people move.
  • Surprisingly litte is known about the role of an individual’s family network in this setting.
  • What is the value of having a family member in a city? Like a person-specific amenity.

Family Networks: Why Do We Care

  1. State of the Art models like Adda, Dustmann + Görlach still need to model other unobserved factors affecting location preference in a random fashion.
  1. How much randomness is left in this model once we account for family network?
  1. Policy: Should we distribute vouchers to people considering moving back to their elderly parents?

Status: early stage, stylized facts, model that runs, lots of parallel computation, threading and multicore.

Not sure it’s the right model.

3. Bus Operator Behaviour In Greater London

Red Buses In London are Privately Operated 😲

UK Competition Commission (CC) 2011

Competition and Markets Authority (previously CC) case

Bus Routes and Garages

Why Does Garage Location Matter?

Why Does Garage Location Matter?

Why Does Garage Location Matter?

  • Location determines competitiveness in bidding for routes.
  • Own and Competitor’s garages!
  • How come?

Dead Miles

  • We find: 13% of London bus drive times are dead

  • Empirically relevant for bidding behaviour.

What about a central Dispatcher?

  • Central planner who decides who has to park bus where.
  • Observing all operational constraints.
  • Reduces dead miles by 14%.

RA Needs

  1. git and github
  2. julia
  3. R: {data.table}, {sf}

+

Basically everything one learns in ScPoProgramming 😉

End

Research Presentation M1 ScPo