If you are a company providing or a municipality supporting on-demand transport services, there is an important decision that you must make: what should be the structure of the service? By structure, I mean how drivers and passengers can interact, or how payments are made. For instance, traditional taxi services differ in their structure from providers such as Uber.
In an earlier working paper (available here) and the post here, I discussed the differences between the main classes of on-demand transport services: hackney carriage; dispatcher; dial-a-ride; and market-based. However, the classification was limited in that it did not distinguish easily between services within a given class.
With the rise in market-based on-demand transport services (e.g., Uber), there is now a need to understand the different ways that these services can operate. And this is what I intend to describe here.
The basic problem in on-demand transport is how to allocate passengers to drivers and figure out how much each journey should cost, accounting for the payment taken by the provider. Through a smartphone app, each market-based service provider is determining the allocation and pricing through a market mechanism. In fact, each of these providers is implementing a two-sided market, where on one side are the passengers and on the other are the drivers.
There are three classes of mechanisms (i.e., ways of allocating and pricing) for two-sided markets: posted-price (see e.g., here); double auction; and hybrid mechanisms. Posted-price mechanisms are currently the most popular approach, and have been adopted by providers such as Uber.
In a posted-price mechanism, passengers and drivers are offered a price for the journey (or payment in the case of drivers), which they can either accept or reject. In practice, this is usually achieved by choosing a subset of drivers that can service each passenger request. A price is then offered to the passenger and a payment is offered to each driver, each of which is free to accept or reject their offer. If the passenger accepts the price, then she is able to select a driver from those that accepted their payment.
Illustration of the posted price mechanism.
Of course, the main problem in the posted-price mechanism is how to set the prices. This is usually commercially sensitive and depends on the data that has been collected by the provider on passenger and driver behavior and spatial-temporal information (i.e., where in the city the passenger is and the time of the request). In terms of the research literature, the price setting is often studied as the multiperiod pricing problem of which there have been several recent contributions (see e.g., here and here). Note that taxi pricing schemes can be viewed as posted price mechanisms and as such there is a wealth of related work in this literature (see e.g., here).
The second class of market-mechanisms are double auctions, which are two-sided version of the auctions that might be familiar from property or art sales (e.g., English or sealed-bid auctions). In these mechanisms, passengers and drivers are able to bid the price/payment they would like. The role of the mechanism is then to be able to match bids from passengers and drivers to ensure that the price bid by the passenger is higher than the payment required by the driver.
Illustration of the double auction mechanism.
There has been limited implementation and research literature on double auction approaches to on-demand transport. The main exception is our paper, which can be found here. As noted in our paper, I believe that the main difficulty in using double auction mechanisms is that they can reinforce existing financial inequality; that is, passengers with significant financial resources can easily use the service, while those with less cannot. The consequence is that the service will not necessarily be used by those that need it most. As such, double auction mechanisms are probably best suited to services aimed at businesses where the issue of financial inequality is less significant.
The third class of market-based approaches in on-demand transport are the hybrid mechanisms, which typically price passengers via a posted price mechanism and set payments for drivers via a single-side auction. At present there has not yet been a published paper dealing with this class of mechanism (although with Nir Oren and Michal Jakob, I have one in the works–a preprint is available here). However, this class of mechanism has been implemented by the provider Liftago in Prague, Czech Republic.
Illustration of a hybrid mechanism.
The main implementation of hybrid mechanisms is used as a way to allocate drivers to passenger requests, one passenger at a time. In particular, for each passenger request, a set of drivers is selected by the provider to bid for the passenger’s journey. After bidding, the payments required by the drivers (perhaps accounting for a subsidy or commission added by the provider) are sent to the passenger, who is free to accept or reject each driver’s bid.
As in the posted-price mechanism and in fact also the double auction mechanisms, there is a need to select which drivers are able to service each passenger. This is known as the market formation problem and approaches must be tailored to the on-demand transport setting. In work with Jan Drchal, Jan Mrkos and Michal Jakob, we are studying the market formation problem for the hybrid mechanism adopted by Liftago using real data from the provider. This work on data-driven market formation problem is discussed here.
Acknowledgements: A big thank you to Emilie Sengelin for the illustrations.
Afterword: I am currently preparing a collection of references for market mechanisms in on-demand transport services here. If you have published a related paper that is not yet added, please contact me at malcolm (dot) egan (at) gmail (dot) com or in the comments below.