Buyer vs Renter — Urban Commuting ABM Dashboard

Simulation Controls

Grid size (G × G)?

100

Number of agents?

100

Number of jobs?

Ownership %?

50%

Job turnover rate?

Move threshold?

Move-decay factor?

0.3

Owner move %?

Sim speed (steps/s)?

Show commute lines Centre-bias heatmap

Legend

Owner (fixed) Renter (mobile) Job Commute path

Theoretical foundations

Spatial mismatch

Affordable housing and employment centres are often misaligned, forcing longer commutes especially for low-wage workers (HUD, 2005). The model lets us observe how this mismatch evolves under different tenure regimes.

Alonso bid-rent & monocentric city

Classic urban economics (Alonso) posits a land-value gradient peaking at the centre, where households trade off commute distance against housing cost. This city is monocentric: jobs spawn with a centre-biased weight w(d) = exp(−d/k), reproducing the rent gradient as an emergent pattern.

Excess commuting & UEI

Hamilton (1982) introduced excess commuting: the gap between observed average commute and the theoretical minimum if every worker took the closest job. Our Urban Efficiency Index rescales this:

UEI = avg(min commute) / avg(actual commute)

UEI = 1 means the city is spatially optimal; lower values indicate wasted travel.

Mobility as urban capital

U.S. data shows ~16% of renters move yearly versus ~5% of homeowners (Gardner, 2022). Mobility lets the labour market re-sort spatially. The model encodes this asymmetry: owners are anchored, renters relocate when commute gain exceeds a threshold.

Commuting & well-being

Long commutes hurt productivity and innovation (Wu, 2021) and reduce life and work satisfaction (Peng et al., 2023). The Satisfaction Index in this model penalises both long commutes and frequent moves — capturing the renter trade-off.

Methodology

Model design

Two-dimensional grid lattice with fixed boundaries (no wrap-around). Centre cells are most attractive; jobs and (indirectly) renters cluster there. Distances are Manhattan: |Δx| + |Δy|.

Agents

Owners hold a fixed home; if their job vanishes they accept the next nearest job from the same address. By default they never relocate. The Owner move % slider allows up to one lifetime move per owner with that probability.

Renters evaluate relocation each step. They move when (current commute − new commute) ≥ threshold × (1 + decay × moves_so_far). Move-decay encodes fatigue: more prior moves → harder to move again.

Jobs

Total job count is held constant. Each step, a fraction (job-turnover rate) of jobs dies and an equal number is reborn at random cells weighted by proximity to centre. When an agent's job dies, they immediately take the nearest available job.

Metrics (the five)

Avg Commute Distance — mean Manhattan distance home→job.
Urban Efficiency Index (UEI) — avg-min over avg-actual commute. Closer to 1 is better.
Economic Productivity — % time saved versus the 100% owner baseline at this configuration.
Agent Satisfaction — per-agent score starting at 100, penalised by commute and prior moves; averaged.
Spatial Efficiency — jobs–housing balance across concentric rings of the grid.

Expected results (paper)

Higher renter share generally → shorter commutes, higher UEI, with diminishing returns. Owner-heavy cities show longer commutes and excess travel.

Satisfaction is expected to peak at an intermediate mix (paper suggests roughly ~70% renters / 30% owners) — enough flexibility to re-sort the labour market without excessive churn.

City lattice

Step 0 · 0 owners · 0 renters · 0 jobs

Performance metrics

Five indices

Avg commute distance

—cells

Urban Efficiency Index

—/ 1.00

Economic productivity Δ

—%

Agent satisfaction

—/ 100

Spatial efficiency

—%

Total renter moves

—

Avg commute distance & UEI

over time

Satisfaction & spatial efficiency

over time

Cumulative renter moves

over time