Economic impact of pedestrianization

Pedestrianized streets have been implemented in the center of various European cities, in the context of sustainable development. Today, under the pandemic of COVID-19, the importance of public spaces is more recognized than ever, as a tool against the infectious disease. However, the impact of pedestrianization of streets rests largely unexplored, when it comes to the economic aspects. Does pedestrianization provide vitality to the neighborhood, or the retail stores along the streets will see a decline of sales volume? This study gives an empirical evidence on these questions.

Methodology

Data retrieval

Street network data is retrieved from Open Street Map, via Ohsome API. In addition to geographical structure of streets, the history of pedestrianization, i.e. when the street was pedestrianized, is obtained. The data is aggregated at grid level, each having the size of 128m x 128m.

Propensity score matching

In order to infer the causal relation between pedestrianization and change in sales volume of retail stores, we first performed propensity score matching. This econometric approach takes variables that may affect the sales volume in a grid, and finds a pair of grids with similar characteristics but for pedestrianization or not.

Difference-in-Differences

Once the grids are grouped into two, namely those with pedestrianized streets and the others, another econometric approach called Difference-in-Differences is employed for causal inference. Under the assumption that the two groups would have performed similarly if pedestrianization did not happen at all, the effect of pedestrianization can be estimated by the following value: (the difference of sales volumes between the two groups after pedestrianization) – (the difference of sales volumes between the two groups before pedestrianization).

In order to robustly check the assumption that the sales volumes of the two groups before pedestrianization behave similarly (= Common Trend Assumption; CTA), lead and lag style regression is adopted. It captures the difference of sales volumes between treatment and control groups for each time period, considering the impact of other covariates. The trend plot thus clarifies whether the CTA holds or not.

General results: different impact depending on the type of neighborhood

Results show that stores located in the pedestrian environment tend to generate higher sales volumes than those located in the non-pedestrian environment. We also found that store density is one of the key factors crucial for this increment, while geographical location of stores does not matter much, suggesting that whether the intervention is implemented in the city center or periphery won’t have differentiated impact on the increment of stores’ revenue. Conversely, stores’ category may have significant impact: while the retail category of ordinary articles is negatively affected, the café or restaurant-related category received a positive impact. We speculate that people might prefer the pedestrian friendly environment rather than the vehicle-oriented environment for those non-tradable local consumption activities.

Publication

Yoshimura, Y., Kumakoshi, Y., Fan, Y., Milardo, S., Koizumi, H., Murillo, J., Santi, P., Zhang, S., Ratti, C (2020) The economic impact by the pedestrianization (in preparation)