Research
“E-hail, flexible fare and leasing: improving New York taxicab market” (Job Market Paper)
Recent literature finds evidence of search frictions in the taxicab market because of spatial mismatches between vacant cabs and waiting passengers (Lagos (2000), Buchholz (2018)). I present a dynamic spatial matching game model to study the effects of matching improvements including e-hail dispatch platforms, and flexible fare and leasing schedules on drivers' income, utilization rates, and consumer surplus for passengers. With the aid of real time price and waiting time data collected from Uber, I estimate the price and waiting time elasticities on the demand side to predict the response on net demand for taxicabs under different regimes. Counterfactual results indicate that using a flexible leasing schedule increases the number of completed trips by 3,710 and improved average shift earnings by $7.2. Waiving the current e-hail booking fee and implementing a universal e-hail dispatcher for the city result in higher usage of e-hail, generating an additional consumer surplus of $0.15 per commuter per day, aggregating to $25.32 million per year. This suggests that operating taxicabs under a large, centralized matching platform network without additional surcharge can be a way to improve the taxicab market.
Dynamic Pricing and Cross-Platform Competition in the NYC For-Hire Vehicle Industry
Dynamic pricing in the for-hire vehicle industry has been shown to filter excess demand and incentivize supply in busy times and locations (Chen & Sheldon (2015), Hall et al (2015)). Even though the taxicab market has begun to adopt centralized dispatch matching (“e-hail"), flexible pricing on e-hail taxi platforms was not approved until August 2018. Dynamic pricing can lead to more efficient allocation of existing drivers as well (Chen, Mislove & Wilson (2015)). I found evidence of spatial reallocation of Uber drivers under price surges in real time data collected from Uber's application programming interface. To study the welfare effects on drivers and passengers under different pricing strategies and industrial organization in the e-hail taxi market, I extend the structural model of taxicab market presented in Wong (2019). Using demand estimates to calibrate the model, simulation experiments suggest that social surplus is highest when the e-hail taxi market is operated by a monopolist. Under all settings considered, surge pricing generates a mutual gain for riders and platform. Regional pricing provides gains for platforms at the expense of passengers except when operated by a social welfare maximizing monopolist.
Work in Progress:
“Seemingly Irrational Serving Strategies: A Dynamic Model of Professional Tennis Players” (with Axel Anderson, Jeremy Rosen and John Rust)
Walker and Wooders (2001) used von Neumann’s Minimax Theorem to model professional tennis players’ serve location strategies. Per minimax, a static point-maximization model has two implications: serves should have equal win probabilities to all attempted locations, and servers’ choices should be serially independent. We scraped data over 850,000 ATP and WTA serves and found contradicting evidence: controlling for server-returner pairs, first serves hit to the returner’s body are 3.4% less successful than non-body first serves. In addition, servers’ choices are negatively serially correlated. We develop a dynamic discrete choice model to study servers’ decisions. In this model, players may hit an inferior serve on the current point or deviate from serial independence to increase their odds of winning future points and ultimately the game. Therefore, common coaching advice to hit body first serves and mix up serve locations may be optimal.
Recent literature finds evidence of search frictions in the taxicab market because of spatial mismatches between vacant cabs and waiting passengers (Lagos (2000), Buchholz (2018)). I present a dynamic spatial matching game model to study the effects of matching improvements including e-hail dispatch platforms, and flexible fare and leasing schedules on drivers' income, utilization rates, and consumer surplus for passengers. With the aid of real time price and waiting time data collected from Uber, I estimate the price and waiting time elasticities on the demand side to predict the response on net demand for taxicabs under different regimes. Counterfactual results indicate that using a flexible leasing schedule increases the number of completed trips by 3,710 and improved average shift earnings by $7.2. Waiving the current e-hail booking fee and implementing a universal e-hail dispatcher for the city result in higher usage of e-hail, generating an additional consumer surplus of $0.15 per commuter per day, aggregating to $25.32 million per year. This suggests that operating taxicabs under a large, centralized matching platform network without additional surcharge can be a way to improve the taxicab market.
Dynamic Pricing and Cross-Platform Competition in the NYC For-Hire Vehicle Industry
Dynamic pricing in the for-hire vehicle industry has been shown to filter excess demand and incentivize supply in busy times and locations (Chen & Sheldon (2015), Hall et al (2015)). Even though the taxicab market has begun to adopt centralized dispatch matching (“e-hail"), flexible pricing on e-hail taxi platforms was not approved until August 2018. Dynamic pricing can lead to more efficient allocation of existing drivers as well (Chen, Mislove & Wilson (2015)). I found evidence of spatial reallocation of Uber drivers under price surges in real time data collected from Uber's application programming interface. To study the welfare effects on drivers and passengers under different pricing strategies and industrial organization in the e-hail taxi market, I extend the structural model of taxicab market presented in Wong (2019). Using demand estimates to calibrate the model, simulation experiments suggest that social surplus is highest when the e-hail taxi market is operated by a monopolist. Under all settings considered, surge pricing generates a mutual gain for riders and platform. Regional pricing provides gains for platforms at the expense of passengers except when operated by a social welfare maximizing monopolist.
Work in Progress:
“Seemingly Irrational Serving Strategies: A Dynamic Model of Professional Tennis Players” (with Axel Anderson, Jeremy Rosen and John Rust)
Walker and Wooders (2001) used von Neumann’s Minimax Theorem to model professional tennis players’ serve location strategies. Per minimax, a static point-maximization model has two implications: serves should have equal win probabilities to all attempted locations, and servers’ choices should be serially independent. We scraped data over 850,000 ATP and WTA serves and found contradicting evidence: controlling for server-returner pairs, first serves hit to the returner’s body are 3.4% less successful than non-body first serves. In addition, servers’ choices are negatively serially correlated. We develop a dynamic discrete choice model to study servers’ decisions. In this model, players may hit an inferior serve on the current point or deviate from serial independence to increase their odds of winning future points and ultimately the game. Therefore, common coaching advice to hit body first serves and mix up serve locations may be optimal.