Spatial simulations

Quantitative models of economic geography are developing quickly (a great overview is here). In the 3rd year bachelor seminar in Urban Port and Transport Economics, we develop one of these models. It contains

  • people's tradeoffs between housing space, consumption goods and commuting costs

  • travel between locations estimated from GIS networks

  • modes choice for commuting (car, public transport)

A new bridge in Rotterdam

The river Maas splits Rotterdam in two: a Northern and a Southern part. The split has important consequences for the city. It poses logistical challenges, as the bridges and tunnels form the few bottleneck to traverse the river. The split is social, too: income, ethnicity and identity are divided by the river.

A new bridge or tunnel is a matter of much debate. The option most discussed is a road over or under the river, connecting de Esch and Fijenoord.

As I was curious about the potential impacts of the bridge, I decided to analyze the bridge in an economic spatial model.

I assumed the bridge would connect to existing road infrastructure. I also assumed that the bridge would relieve 30% of the congestion of nearby bridges. I entered the bridge in my model, which includes:

  • two groups: households of low income and households of high income (separated at 40k euros/year)

  • the wage faced in local labor markets very by group

  • transport choices are income-specific

  • different experiences of quality of life across the groups for residential areas (for instance, income determines the wishes for restaurants, stores, nightlife or parks in the area)

I used the TOPNL infrastructure files from PDOK, MON/OVIN mobility surveys (available after request) and CBS shapefiles and neighborhood information. In the analysis, I formed areas based on the first three digits of every postcode.

relative pop growth by pc3
relative job growth by pc3

What are the main results? There is a slight population increase on the Southern bank near the bridge, and a slight decline North. Somewhat further away, we see larger increases in population. Jobs change more strongly: the center of Rotterdam offers employment opportunities, the bridge allows people to travel more easily across the river, opening up more jobs. In part, the bridge seems to enable people to work in the center or Rotterdam, but live outside the city.

The changes that the model predicts are small. The reason is that although the new bridge affects travel times, its impact is limited. The congestion in the bridges itself is not huge, and for many people, the new bridge will not change their fastest route much.

The bridge also connects two parts of the city that are very different: one side is historically richer than the other. So what happens to the mix of low and high income households across Rotterdam, after the bridge is built? The city center and the North are home to relatively fewer rich households, but areas outside the city (in particular south, crossing the river Oude Maas) see more rich households. Areas to the northwest host more lower income residents. The labor market south of the river Maas sees relatively many workers from low-income households, while the labor markets north of the river show more high-income jobs.

What could explain this pattern? From the observed commuting patterns, the labor markets just south of the river are particularly appealing to low-income workers, and increasing access to such areas makes them concentrate further. On the northern side, labor market attractiveness is more balanced between low and high income workers, and increasing access does less to change the balance of low and high income workers. Similarly, areas outside the city center are comparatively attractive to high-income workers, and the bridge allows them to extend the range of their commute. However, there is an important additional reason: the models shows that high income workers are less deterred by distance in their commuting choices, so they are more likely to exploit reduced travel times to move to homes and jobs further away from the city.

In short: employment concentration of lower income workers in the center south of the river, residential movement of high-income households out of the center, mostly south and east.

A few grains of salt come with this exercise. First, many motives to seek a location to live or work are not included in the model. For instance, mixing or segregation is in people's residential preferences. Adding more people to a local labor market could lead to agglomeration, congestion and composition effects that are not accounted for. And, if these people start to live somewhere else, specific roads and transport options might congest, while other become less congested - that is ignored too. The bridge could have symbolic meaning, as the Erasmus bridge does - all neglected here. Second, this model is valid for small shocks. If the shocks are larger, less expected things will probably happen: there are tipping points in gentrification, multinational headquarters might settle in the area, or new industries emerge.