Didi Chuxing, the biggest ride-hailing company in China, said the company lost 2% of the fare for every ride it operated in China for the three months ended in December 2018, in an essay published on April 22, 2019. It was the first time the company published such metrics.

DiDi (download the 44-page report about DiDi for free) took 19% in commissions on average for the actual fare riders paid, or so-called take rate. Drivers took all the long-distance dispatching fee, dynamic fare, and tips, which are not included in commission calculation. Didi’s executive CHEN Xi (陈熙), who oversees the company’s ride-hailing business, said that the take rate changes dynamically.

A sum equivalent to 10% of the fare is used for operational costs, which includes technology research, safety, calling center, human resource, and offline operations. The company wants to cut expenses while staying a low gross margin. A further 7% goes to the driver as incentives to award drivers who service in rush hour and hot spot district. 4% is used to pay taxes and online payment-associated fees. In conclusion, while Didi took 19% in commissions on average, its costs surpassed that at 21% of the fare. Didi had to cover the rest of 2% costs from its fundraising.

CHEN mentioned that it is a normal phenomenon in the ride-hailing industry. Uber’s take rate in its core ride-hailing platform was 20% in 2018. Its smaller rival Lyft registered 26.8% take rate in 2018, according to its prospectus. 

DiDi and Uber have played in the first half game in the ride-hailing industry in China. DiDi achieved a monopoly position in terms of traffic flow after it had won the battles with Kuaidi and Uber. In 2018 Q2, DiDi was accounted for nearly 96% of the market share, which was significantly higher than the second place Shouqi. In the second half game, however, regarding the B2C model, there is no monopoly like DiDi.

However, DiDi is now being trapped in a dilemma: the company will have face losing market share or losing money. The company cannot increase its take rate or cut driver incentives to offset losses. If that happens, drivers will flow to other Apps easily. The company cannot cut customer subsidies completely either for this will also lead to customers using other ride-hailing Apps. We believe the main reasons are two: the Asymptotic Value Marketplace and low switching costs. 

Two-sided network effect

Network effect occurs when a company’s product or service becomes more valuable as usage increases. However, it can be divided into two types: two-sided network effect and one-sided network effect.

One-sided network effect is also known as same-side effect. As users increase, so does the value. It happens mostly on social networks like WeChat.

Two-sided network effect. More users increases the value of the services they receive from the other side and vice versa.

DiDi has created a virtuous two-sided network effect: a large supply of drivers enables more rides to be completed, which attracts more riders, making the entire platform more valuable for both sides. In essence, each incremental rider and driver creates a network effect. Driven by its strong two-sided network effect, DiDi has built three pillars: maximum supply liquidity, minimum wait time, and seamless service. However, DiDi does not perform the one-sided network effect.

For the first part of the network effect: the core mission for DiDi is to acquire enough users or riders. DiDi has cooperated with WeChat and Alipay to put an entrance on their APPs. By pouring money into marketing and subsidizing, it can attract more new users. DiDi needs to improve its customer loyalty to retain new users to keep on using its platform.

Second, dynamic pricing. The dynamic price adjustment ensures a reliable service for riders and incentivizes drivers to provide rides at peak times and location. Efficient dynamic prices depend on stock of drivers elasticity, according to Jonathan Levin and Andy Skrzypacz. In addition, Dynamic Pricing is particularly important for ride-sharing platforms because pricing too low for the market conditions creates a "wild goose chase" phenomenon: demand outstrips supply, and pickup ETAs (Estimated Time of Arrivals) suffer long waiting time, according to Dawn Woodard.

DiDi introduced Dynamic Pricing on December 2018 to Australia to help boost drivers’ earnings. During times of high demand, Dynamic Pricing will apply on drivers’ trip fares. If demand is high in a particular area and time, the trip fare will have a multiplier added to the base, waiting and distance fees.

We define take rate as ride-hailing net revenue as a percentage of gross bookings. It fluctuates based on competitive pressure, the dynamics within each segment market and product mix. It affects profits. The firm will adjust its commission fee and service fee accordingly to match both needs of drivers and riders. Ride-hailing companies take a 20%-25% cut of drivers’ fares (gross bookings). On the other hand, ride hailers need to recruit new drivers to grow and need to subsidize (or even occasionally fully pay) those drivers to attract new customers. This is the main reason why no ride hailer has ever booked positive operating profit to date. Both the service fee paid by drivers and the driver incentives affect the take rate, which in turn affects the net revenue. Uber’s core platform take rate was 20% in 2018.

DiDi’s network effect benefits both drivers and riders and creates a continuous cycle. By focusing on a small number of locations at first and gaining reputation via word-of-mouth marketing, DiDi soon expanded to 400 Chinese cities by March 2017 and 1,000 cities globally by Jan 2019. With scale, DiDi has attracted enough backers to support its generous subsidy plan, which allows the company to continue scaling regardless of costs.

The Asymptotic Value Marketplace

DiDi only has a strong two-sided network effect but weak one-sided network effect. DiDi intended to build the one-sided network effect by adding a social network function in DiDi Hitch, a carpooling service. This function allowed drivers to create public tags for passengers including physical features, gender and age. A female passenger using the service was killed in May 2018 and the public criticized DiDi’s efforts to market Hitch as a “social carpooling” service. As a result, the company removed social features from Hitch service. 

We believe Didi has a strong two-sided network effect. However, not all two-sided marketplaces are the same. One way to differentiate them is to draw a value-to-demand curve. This value curve refers to how fast the value to the demand side increases as supply increases. The graph above represents the Typical value marketplace, the Delayed value marketplace, and the Asymptotic value marketplace.For Asymptotic Value Marketplace, the initial supply quickly adds value to the demand-side, but soon the value to demand-side growth rate starts to diminish.

Delayed Value Marketplace: In this case, players need to grow the supply-side to a very high level before there is any value to the demand-side. For example, Taobao, JD.com, and Meituan have a Delayed Value Marketplace because the value generated to customers grows slower than the supply. However, once the Critical Mass was achieved, the network effect becomes powerful and value rises rapidly.

The Critical Mass is critical. If we consider the value to demand as the minimum wait time, DiDi’s situation looks like this:

Up to the Critical Mass, DiDi will increase its supply of drivers, leading to a higher capacity per driver and lower idle time for riders. Riders are willing to pay enough money for DiDi cutting their wait time from 8 minutes to 4 minutes.

Beyond the Critical Mass point, however, the value diminishes. Demand utility will not improve as fast as before for there is less difference between waiting for 4 minutes and 2 minutes. DiDi will then have to face diminishing returns to scale. 

Vulnerable to competition. Asymptotic Marketplace is more vulnerable to competition compared to the Delayed Marketplace. Competitors can easily enter the market as long as they can reach a sufficient threshold to provide comparable pickup times or high-priced and service-quality premium car service. That is part of the reason why Meituan ride-hailing service, Shenzhou Zhuanche, Shouqi Yueche, and Caocao Zhuanche can impact on the company’s business. 

Low switching costs

Ride-hailing is a type of service or commodity sending people from location A to location B. For users, they can download different ride-hailing Apps and decide which has the best price/service without recognizable costs. For drivers, things are pretty much the same. Both sides of the marketplace lead to low switching costs. It’s also called “multi-tenanting” when supply or demand is willing to live in two or more places.

We took the battle between DiDi and Meituan as an example. In January 2018, DiDi unveiled its bike sharing platform, which integrates services of two Chinese bike sharing companies OfO and Bluegogo and is open to other potential partners in the future. The battle between DiDi and Meituan has escalated in 2018 when Meituan bought bike-rental startup Mobike after DiDi launched its bike-sharing business. Meanwhile, DiDi made a move into food-delivery industry.

DiDi’s market share rebounded after Meituan’s subsidies canceled. Meituan Dianping (3690.HK), the operator of the world’s largest on-demand food delivery service, introduced its ride-hailing operation in February 2018 in Nanjing and rolled out the service in Shanghai in March 2018. The cost of rides hailed on Meituan started at CNY 5 (USD 0.75), compared with the initial charge of CNY 14 on a regular taxi. Meituan gave subsidies to drivers as well.

In May 2018, local authorities criticized this kind of price war as unfair competition. Meituan’s subsidies ended, which led to its respective market share decreasing from both drivers side and customers side. DiDi’s market position rebounded as a result. It reflects low switching costs and limited short-term customer loyalty on DiDi’s platform.

Meituan halted the services in September because DiDi’s safety issue has changed the market dynamics.

Before the Critical Mass we mentioned in Asymptotic Value Marketplace, riders’ utility will be improved drastically when DiDi provides more drivers and better technology. Their rides can arrive in less than four minutes. After that, however, DiDi will find out that even with more cars waiting on the platform, it becomes harder to comprise the ride-hailing time to less than two minutes, due to the real traffic situation.

Several weak including Shouqi Yueche, Caocao Zhuanche and Shenzhou Zhuanche get popular as DiDi is under government's strict supervision and compliance pressure. For instance, Shouqi relies on its regional advantages in Beijing, targeting high-end mobility demand, cutting into the car segment market with standardized services, and creating a small and beautiful high-end car-hailing platform. It registered 4.3 million MAU and 665,000 DAU in December 2018, driven by a continuously growing user base.

At present, DiDi is a leading company among enterprises that adopt C2C (Customer-to-Customer) business model, which are typical with light asset. 

We come to the conclusion that DiDi has a stable moat. Without disruptive technology (for instance, Alphabet’s self-driving unit Waymo ventures into ride-hailing with self-driving technology) and extreme political factors, it is hard to find a new company that will threaten DiDi’s business.

However, we think that DiDi is exposed to frequent risks while the shortage of transportation capacity continues to plague DiDi in the short term.