Predicting how groups of people will behave in the future is complicated, so naturally, finding an accurate method to measure how frequently someone will use a micromobility vehicle is complex. For decades, governments have resorted to experts in economics, public policy, urban planning, and transportation to find a solution to best forecast how many resources to allocate for public transportation services and infrastructure based on demand and human behavior. With the availability of real-time data and the rise of big data analysis, governments can now better predict behavior patterns and design more effective transportation systems.
As cities welcome new mobility options, they are also apprehensive of the quasi-novel business model employed by the micromobility companies entering cities across the world. Consequently, many governments have rolled out relatively conservative policies and guidelines for micromobility companies participating in the sharing economy. Most of these guidelines include a maximum fleet capacity and the rationale behind that cap. Using real-time data cities can avoid over- or under-estimating demand and adopt a more flexible control. In order to contribute to that real-time data analysis Grow proposed a dynamic occupancy index to improve the distribution of the scooters in the public space.

Examples of cap regulations in micromobility

In Latin America, some cities are employing a hard cap on the number of micromobility vehicles allowed for deployment. For example, as part of Mexico City’s permit program, the city implemented a maximum quota of 3,500 scooters across all companies, with additional limits set on the number of vehicles per provider. Each company participated in a lottery and auction system where they could compete to win the highest number of vehicles allowed. The city used an algorithm to determine the fleet size of each company. With this system, Grin Scooters gained a permit for 1,750 scooters. Based on historical data, before this cap was implemented Mexico City had over 7,000 scooters from Grin, Lime, Bird, and Movo and their service areas covered greater parts of the city. This allowed scooter sharing services to reach a broader community by ensuring scooters were always available to cover last-mile trips in more neighborhoods than they do now with the new permit system. With limited scooter availability and service areas, the cap on scooters created a negative incentive for users since they now have to walk longer distances to find an available scooter or end their rides before reaching their destination.

Similarly, Bogota’s local government rolled out a permit that included a cap on the fleet size of micromobility companies. The city landed on 3,050 scooters as the total number allowed on public spaces and to be divided equally with all the companies applying for the permit program. As of now, Lime, Movo, Muvo, and Grin applied for the permit and each is allowed to operate up to 712 scooters. Conversely to the cap set in Mexico City, Bogota’s permit does allow for the operation of a larger scooter fleet as long as the vehicles are parked in private spaces or outside commercial zoning areas. Despite this fact, the set cap is detrimental to the effectiveness of scooter-share services aiming to increase the city’s quality of life by reducing car use and fomenting an efficiently interconnected public transportation system.

Adding a vehicle cap results in problems that lead to inefficient practices that deteriorate user experience across mobility platforms. Governments can request support from micromobility services to combine and analyze all mobility and transportation data which will allow cities to better determine the needs of the community.

Density Analysis Model

Nonetheless, utilizing real-time data produced by the users of micromobility services can increase the accuracy of a vehicle capacity forecasting model.
In an attempt to strategically predict the short- and medium-term supply needed in each city, at Grow Mobility we developed a density analysis model that is run monthly and which generates an estimate for the number of vehicles per kilometer squared in a given city. For this model, all of Grow’s coverage areas and the number of active vehicles are measured and recorded. A coverage area is defined as the operating polygon in a city where users can find and use Grow’s vehicles (See Image 1 for example). The unit used for these polygons is ‘km^2’.

Source: Google Maps

Moreover, active vehicles are defined as the number of vehicles on the street that had a ride during a given period. Specifically for the vehicle density analysis, the average active vehicles used comes from the most recent week (last 7 days). Once the quantities for coverage areas and average active vehicles in each city are gathered, the density can be calculated. The units used for vehicle density in this model are ‘vehicles/km^2’.

To visualize how each city compares with one another, the densities are placed on a box plot. The image below shows the most recent update of the density box plot for Grow’s scooters.

Self elaboration Source: Grow Inc. 

The vehicle density changes in every city since the behavior of users is linked to factors that change city to city so clear conclusions can be made based on the profiles of Latin American cities. For example, Rio de Janeiro continuously has one of the highest densities but Cali has one of the smallest densities. On the one hand, Rio de Janeiro is a large, tourist-heavy metropolis, making its market unique from other cities since potential users include local commuters and temporary visitors. Small-sized cities with limited tourism tend to have smaller vehicle densities since the market requires less micromobility vehicles to meet the local demand.

The same exercise can be replicated for other companies using scooter fleet size estimates. This exercise helps create benchmarks for Grow by comparing density to other companies in cities in Latin America, Europe, the US, and New Zealand. Chart 2 illustrates the densities for all the participating companies in the chosen cities. Most non-Latin American cities tend to have smaller density levels while Latin American cities tend to have larger densities. The size and structure of the service areas can partially explain the differences between the regions. Service areas in cities like Mexico City or Bogota are limited by topographical barriers and safety. Offering micromobility services in most Latin American cities for an additional street could mean putting users at risk of injury or inside a crime-heavy neighborhood.

Self elaboration Source: Grow Inc. 

Estimating Scooter Capacity and Demand

Along with Mexico City and São Paulo’s scooter densities, the highest, lowest, median and weighted (Active scooters are used as the weights to get the ‘weighted average’) average densities are used to estimate various levels of scooter capacities. This is done by multiplying the size of all the service areas by the selected density. For example, the weighted average density for the most recent update to the analysis was 37.6 scooters/km^2 and the sum of the service areas was 583 km^2. After multiplying these quantities, the estimated scooter capacity is 21,943 scooters if all the cities had the same density as the weighted average. The same calculation can be repeated for specific cities or countries.

For instance, by using the weighted average density and its current service area size, Santiago, Chile could have up to 3,122 Grin scooters. Chart 3 shows the latest results using different levels of density (scooters/km^2) to get an estimate of the scooter capacity per country.

Self elaboration Source: Grow Inc. 

Every city is unique so some cities could have as many scooters as the ‘Highest’ scooter density level and other cities will only be able to reach the ‘Median’ or ‘Lowest’ density levels. Vehicle density is a good starting point to figure out the demand for micromobility vehicles, but other measures and strategies should be used to achieve the best demand forecast. The utilization rate (rides per scooters) of each city can be used as a supplement to support the case for increasing or decreasing fleet size. The utilization rate is a helpful metric since it demonstrates if a city is performing well or not. A low utilization rate can signal low demand, so when the utilization rate increases it indicates that the demand is increasing. Cities could set a ‘floor’ for the number of vehicles each micromobility company has and use the density and utilization data produced by the companies to determine whether or not the public demands more vehicles on the streets.

Density Model Limitations

This density model does have some constraints since it only uses certain variables without considering other factors or externalities. The model uses active scooters, or scooters with a ride during the chosen period, and disregarding scooters on the streets that failed to get a ride. Doing this prevents the over- or under-estimation of the vehicles used, but the inactive vehicles can also be assumed to be “littering” sidewalks without movement. Despite this, it is in the best interest of Grow to reduce the number of inactive scooters by perfecting vehicle placement and the size of its fleet as issued in past blog entrances. Unused scooters are not generating revenue, so Grow will naturally reduce fleet size in a city where the utilization rate is too low. Another limitation of this model is the use of data with restricting regulations set into place. These regulations can affect the number of vehicles available or service area sizes. City-specific regulations must be considered when interpreting the density since some cities could have many different results if arbitrary regulations were not put into place.

Using vehicle density can help produce an ideal environment for the local market to determine the quantity of micromobility vehicles needed to maximize utility in the community. A robust model to measure the demand for micromobility services must include real-time data to prevent local governments and special-interest groups in mobility from making erroneous assumptions. When cities set caps or quotas on the number of deployed vehicles, they are creating anti-competitive barriers that affect the quality of life for city dwellers. Strict regulations that limit availability and choice deprive groups of people access to micromobility services. Additionally, such regulations provide a competitive advantage for wealthier companies, restricting the ability of smaller companies to participate in the market. Ultimately, cities obliquely incentivize car use by setting caps on the availability of micromobility vehicles.

Alex Rios is the author of this article. She is the Head of City & Competitive Intelligence at Grow Mobility, where she analyses cities and micromobility around the world.