We tend not to like competition – if we’re the provider of whatever it is. But there are times when having people come into the market to compete with you is beneficial. Say, in bike sharing. Where the concept is pretty new, habits aren’t as yet formed. So, someone else investing in converting behaviour has spill over effects onto a rise in your own sales:
Market expanding versus market stealing: Competition with network effects in bike-sharing
Guangyu Cao, Ginger Jin, Xi Weng, Li-An Zhou 09 November 2018
Positive network effects may lead to winner-takes-all in some markets. The column analyses dockless bike-sharing in China to show instead how an incumbent can benefit from positive spillovers from a competitor’s entry. In the case of bike-sharing, consumers multi-home, the market exhibits positive network effects, and investment by two firms is more cost-efficient than investment by one.
Competition in network markets has caught the attention of researchers and policymakers. Positive network effects may lead to winner-takes-all, but this anti-competitive concern may be reduced by multi-homing and compatibility (Katz and Shapiro 1985, Farrell and Klemperer 2007).
Positive network effects and multi-homing users change how firms compete, but it is not clear how incumbents should change prices, sales, and investment when competitors enter a market, or the extent that a competitor expands or steals the user base of the incumbent. There may also be other competitive considerations as well as winner-takes-all. In a recent working paper (Cao et al. 2018), we study these issues in the context of bike-sharing.
The rise of dockless bike-sharing
Bike-sharing systems have gone through many generations since the 1960s, mostly driven by technological development (DeMaio 2003, DeMaio and Gifford 2004, DeMaio 2009). In a traditional system each bike is docked at a station, riders must pick up the bike from one station and return it to this or another station of the same network. Docking stations may not be near either origins or destinations, and the capacity of stations is limited. Dockless bike-sharing systems reduce these problems. In a dockless system, bikes can be parked freely on the pavement, within city-authorised areas. Users scan the QR code on the bike’s smart lock using a smartphone and reset the lock at the end of the trip. A slogan coined by ofo, one of the market leaders, claims that dockless bikes can be used “anytime and anywhere”.
Our study focuses on ofo and Mobike, the other leading firm in dockless bike-sharing. Both originated in China but now operate worldwide. ofo was the first dockless system, and it launched on 7 September 2015 in Beijing. Its bikes are coloured yellow. Mobike started in Shanghai on 22 April 2016 with orange bikes. Industry reports find that ofo and Mobike accounted for between 90% and 95% of the Chinese bike-sharing market from the beginning, and most users multi-home.
Does entry help or hurt the incumbent?
Using news reports and ofo’s internal data (up to 14September 2017), we identify 59 cities that were first served by ofo, with Mobike entering the market later.1 We label these ‘ofo first’cities. There are another 23 ‘ofo alone’ cities and 22 ‘Mobike first’cities. Figure 1 shows the geographical distribution of these cities. Taking ofo as the incumbent and Mobike as an entrant, we applied difference-in-differences to the sample of ‘ofo alone’ and ‘ofo first’, defining Mobike’s city-specific entry as the treatment.
Figure 1 Geographical distribution of different city groups
We found that Mobike’s entry boosted ofo’s daily trip volume by 40.8%. It also increased ofo’s average revenue per trip by 0.041 renminbi, which suggests that the increased trip volume was not driven by an intense price war.
Omitted variables or endogenous entry may still challenge the causal interpretation. To address this, we confirm that ‘ofo first’ and ‘ofo alone’ cities are comparable in pre-treatment trends after extensive controls, and the baseline results are robust to heterogeneous time trends, placebo tests and subsample regressions. We also predict Mobike’s entry date in a city using the timing of Mobike’s venture capital funding (eight rounds in total), and the city’s existing attributes such as population, seasonality, and transport infrastructure. Then, using the predicted entry date as the instrumental variable for the actual entry date, we confirm that Mobike’s entry benefits ofo in both trip volume and revenue per trip.
Market expanding or market stealing?
Usually, an entrant steals consumers from the incumbent and presses the incumbent to lower its price, unless the entry expands the market demand. But even if the overall market expands, there should be market stealing between Mobike and ofo, as most consumers multi-home. Indeed, Mobike’s entry reduced the percent of old users that remain active on ofo, but this market-stealing effect is dominated by expansion in new users. Consistently, in the days immediately after Mobike’s entry but before ofo made any new bike investment, we find that ofo lost old users and did not pick up extra new users.
Bike investment and bike utilisation
Data suggest that ofo has put more bikes in the ‘ofo first’markets since Mobike’s entry, above and beyond the periodical bike investment it made in ‘ofo alone’ markets. This could explain part of the market expansion effect, but it does not explain why ofo’s bike utilisation rate – measured by the number of trips per ofo bike per day – did not increase significantly on Mobike’s entry. Furthermore, Mobike’s entry allows ofo bikes to reach more grids2 and become more dispersed in the city.
All these results point to a competition-reinforced network effect. It is likely that Mobike’s entry has expanded the overall network of bike-sharing, attracted more consumers to join bike-sharing, and helped to expand the ofo network in depth, width, flatness and user reach.
Putting all elements into a theoretical model
To best understand the empirical facts, we write a model with one or two firms in the market. In both types of market, consumers decide whether to search for a bike, given the price of each firm and the expected probability of finding any bike. Firms decide on price and bike investment, which influences the matching probability and is subject to convex costs. Congestion creates a negative network effect, but if the matching technology exhibits increasing return to scale, it also creates positive network effects. Give these network effects, we derive how firms choose price and bike investment in equilibrium.
The model predicts higher price in duopoly than in monopoly, because each searching consumer searches once and will take the found bike no matter which firm it belongs to. The monopolist must fully incorporate the negative impact of raising prices on consumer willingness to search, but a duopolist ignores the negative spillover of its own price hike on the competitor. This incentive leads to higher price in duopoly.
In contrast, the comparison on trade volume, bike investment, and bike utilisation rate depends on the network effects. When there is significant increasing return to scale in the matching technology, competition could generate a market-expansion effect that is large enough to dominate the market-stealing effect. In that case, each duopolist has an incentive to make greater bike investment than the monopolist, and as a result serve more trips and enjoy a higher bike utilisation rate.
Put another way, market expansion occurs not only because of multi-homing and compatibility, but also because bike-sharing has positive network effects. A user who rides a bike from A to B makes the bike available for the next rider at B. This ‘consumption as supply’ could make the matching more efficient as the network size increases. When thousands of consumers use bike-sharing in a small area, the wide availability of bikes increases the expected probability that customers will find bikes at the times and places they need them. More efficient matching also motivates firms to put more bikes on the market, which further increases the demand.
Nevertheless, these network effects must be traded off against the cost of bike investment and maintenance. If the investment cost was convex in the number of bikes, it could constrain the network size from infinity, and therefore leave room for competitive entry.
Our model predicts that when the positive network effects are sufficiently large but not too large to overcome the convex investment cost, each duopolist could engage in more bike investment than the monopolist. This is because each duopolist free-rides on the competitor’s investment. Every bike invested by firm 2 costs nothing to firm 1, but it expands the overall market and benefits firm 1. Moreover, with positive network effects in the overall matching technology, firm 2’s investment makes firm 1’s investment more efficient in attracting users and improving the matching rate. Monopoly cannot achieve the same efficiency because the monopolist must invest at its own cost to get to the same scale and that cost might be too convex to justify the investment.
Implications for public policy and future research
Our work has two policy implications. First, it challenges winner-takes-all in a nascent market with positive network effects. According to that concern, positive network effects would enable the incumbent to become a natural monopoly and then abuse its monopoly power to harm consumers. In bike-sharing, we show that the incumbent could benefit from a competitor’s entry even if their goods are close substitutes. The entry creates positive spillovers to the incumbent because consumers multi-home, the market exhibits positive network effects, and investment by two firms is more cost-efficient than investment by one.
Our results also highlight competition with the outside good. Many theories of network markets, such as Armstrong (2006) and Bryan and Gans (2018), emphasise head-to-head competition between firms but assume away their competition with the outside good. In our context, attracting new users to search for a bike is essential to market expansion.