Some Urban News

I just discovered a massive magnum opus, the Urban Simulation platform UrbanSim, open source on GitHub.

UrbanSim ( is a model system for analyzing urban development. It is an open source platform that has been continuously refined and distributed for planning applications around the world for over 15 years.

There are a number of corresponding academic papers, for instance here.

Some recent news:

Is homelessness a symptom of urban failures or is it, on the contrary, evidence of the attractiveness of urban areas for the country’s poorest?

Lots of experiments in Canadian mortgage markets –> should be a source of natural experiments for research projects.

How technology actually affects a still pretty traditional industry.


Econ 101: Subsidies and taxes aren’t paid by who you think

Our good friend Ronan Lyons from Trinity College Dublin is interviewed in an Irish article, titled “House price increases set to overtake grant offered by Help-to-Buy Scheme”:

The average price paid for residential property increased by €18,954 in the last 12 months, according to the latest Property Price Register.

The rising cost of buying a home has effectively almost wiped out the benefits of the Government’s Help-to-Buy Scheme, which gives first-time buyers up to €20,000 towards the cost of a property.

The average price paid for a house or apartment nationwide is now €256,193 – an increase of 8% when compared to the same time last year, but well short of the numbers though necessary to meet housing needs.

Well, this goes back to a basic principle of public economics: with an inelastic supply (and Ronan says Ireland needs roughly three times as many new homes to be built per year as is currently the case), a subsidy to buyers will lead to a price increase that could offset most of the impact of the subsidy. Simply said, that subsidy is likely a subsidy to homeowners, and not so much a subsidy to buyers. Now Econ 101 does not predict that (in partial equilibrium) the price increase caused by the subsidy will be higher than the amount of the subsidy, and there are other market forces at play here.

A good assignment for the applied economist is to compute the deadweight loss using estimates of housing supply and housing demand elasticities.

Link to the article.


Upward Sloping Demand Curves and Museum Congestion

In a recent post, Matt Kahn revives a question that I get in every first session of an MBA class. Can demand curves be upward sloping?

And in his post:

A museum (as I learned at Versaiilles, France recently) suffers from congestion problems.   This congestion is especially bad if the price of entrance is $0. (…) BLP should reunite to write a structural IO paper on museum demand and the resulting consumer surplus with  endogenous product attributes!

I am particularly interested in BLP models of consumer choice, and there is a paper on the topic (Maddison and Foster, 2003), which studies congestion costs at the British Museum:

The paper does not per se execute Matt’s idea. The paper shows visitors to the British Museum pictures with different levels of congestion and asks them whether they would prefer free admission or 3 GBP (resp. 8 pounds, etc). There are some significant issues with an approach of this kind (see Do People Mean What They Say? by Bertrand and Mullainathan 2003), but the results are interesting. A more ambitious study would randomize actually paid prices (as in Levitt’s Uber paper) and then estimate demand, while instrumenting for congestion (i.e. using idiosyncratic sources of variation for congestion, weather, or other sources). There are many ways in which IO can push the frontier in the estimation of consumer demand.


Johannesburg in 2017

My wife and I spent about a week driving around the city of Johannesburg. While crime statistics in South Africa can be scary for the casual traveller, I found that the wealth of culture, experiences, languages, and ethnicities was well worth the effort. This is a city with all of the 11 official languages of South Africa! This is also a city with large socio-economic disparities that have not significantly been offset by the end of Apartheid. A shocking statistic coming from the Census is that the ratio of White to Black income barely changed over the last 30 years.

In the city of Johannesburg itself, the geography of the city center has been dramatically transformed over the last 3 decades. Business activity has largely moved out of downtown Johannesburg, where the imposing facades of Anglo American’s offices are still dominating the landscape. As in a city like Detroit, the scale of architecture in downtown does not seem to match its current purpose. Large parts of Johannesburg business have moved to Sandton, where glittering buildings, manicured lawns, and large security details fence areas from the surrounding activity. In a perhaps ironic twist of history, Sandton’s statue of Nelson Mandela sits at the center of a high-end luxury mall. While World Bank statistics suggest that about 17% of the population lives below the poverty line, and the Gini stands at a very high 63.4, one sure thing is that such high-end malls see a mix all ethnicities and a thriving (but likely small) Black upper-middle class.

South African data is good. My World Bank contacts are fascinated by the care and the precision of its statistical agency when it comes to collecting census data. So I downloaded the Census of Communities 2011 to draw a map of average household income by ward.

The distribution of household income suggests that downtown Johannesburg is indeed the poorest ward of the city. While high-income households are not living in downtown — and have likely moved out of it –, they have not moved to the farthest wards of the city. In fact they have moved to suburban areas south, east, and north-west of downtown, within reach of downtown. As in the U.S. suburbs of Johannesburg are well-connected with a strong system of highways, and feature malls and other shopping facilities.

Notes: log income at 11.5 is about 100,000 South African Rands, or 7,388 USD at the current exchange rate. This is about 20-25% above the country’s GDP per capita. Household income is reported as a count of households per bracket. See the questionnaire here. Average income is proxied by the average of the brackets’ mid point. The white areas are areas with no reported household data.