into account the demand and supply of rides,” said the young man.
“Oh wait . . . Go slow please. You’re talking to an ‘uncle’ who has only ‘O’ level, OK? I’ve no research experience at all,” laughed Teng while acknowledging his limited formal education.
“Sorry, Uncle. Let me explain slowly. Grab’s prices are determined by supply and demand. It is not fixed by the meter. In other words, it’s dynamic or flexible, depending on the situation. When demand is very high, the fare can go up a lot, much higher than the metered taxi fare.
“Let’s call this the surge factor. I know it’s a big word. But bear with me. The surge factor represents the relative price difference between the fares charged using a ride-hailing app and a standard taxi,” explained the young man slowly.
“For instance, if the Grab fare is $10 and the taxi fare is $9.50, then the surge factor is low because the price difference is small. But if the Grab fare is $15 and the taxi fare is $9.50, then the surge factor is high because the price differential is large.
The surge factor represents the relative price difference between the fares charged using a ride-hailing app and a standard taxi.
“So, why would Grab fare be so high, you might ask. That happens when there is more demand than supply for Grab. This would be the case when it is raining heavily and many people need a ride than there is supply of Grab cars or taxis.”
“Or when the MRT breaks down,” interjected Teng as he recalled the numerous MRT (Mass Rapid Transit) breakdowns that Singaporeans have experienced.
“But for taxis,” the young man continued, “the fare is based on distance travelled as measured by the meter rather than on demand. This results in Grab prices possibly becoming higher than taxi prices. When Grab prices are higher than taxi prices, the surge factor is more than one.
“Is this OK so far?”
“Eh . . . Yes. I think I understand,” replied Teng tentatively.
“Now, here’s the reverse. When there is more supply of Grab cars than there is demand, that is, there are more non-hired Grab cars on the road than there are people wanting a ride, it is likely that the Grab fare will be cheaper than the metered taxi fare. If that is the case, the surge factor is equal to or less than one.”
“OK. I think I follow so far. Surge factor is greater than one when Grab fare is more than taxi fare. Surge factor is one or less when Grab fare is less than taxi fare,” clarified Teng just so that he got it correct.
The young man nodded.
“But please, don’t get more technical. Otherwise, I think I’ll be lost,” pleaded Teng.
“Don’t worry, Uncle. You are doing fine. We’ll take it slowly,” smiled the young man encouragingly.
“My professor used two sets of data. The first dataset is on taxi mobility. This contains the number of taxi trips across all operators and the trip types at every 30-minute interval. This means we know where each trip started and where it ended. We also know whether the ride was initiated through a booking, a street pick-up, a limo service or whatever.
“Because my professor wanted to find out the effect of surge pricing on taxi supply, he focused on taxi bookings of the two taxi operators who at that time forbade their drivers from participating in the Grab service. By doing that, he can tell the effect of surge pricing on the availability of taxis from these two operators in a particular area.
“His data were collected every 30 minutes. Do you know why? Because that would allow us to see whether there are differences during peak hours and end-of-school hours versus non-peak hours. Is this good?”
“Yup, I got it so far. How about the second dataset?” enquired Teng.
“The second dataset contains a panel of all taxis in Singapore. We also collected the data at every 30-minute interval. Here, we know for each taxi whether the taxi is available or hired, and its location. This allows us to understand the supply of taxis. ‘Available’ taxis means they are available for hire, which means there is excess supply since they are not used; while ‘Hired’ taxis means they have been booked already by demand. And we do this for each operator in each area every 30 minutes. Understand?”
“Got it!” said an excited Teng.
The young man continued. “So we know the supply of taxis. How about Grab’s supply of cars? Grab offers a range of ride services – GrabHitch, JustGrab and so on. We studied JustGrab. JustGrab allows commuters to book a ride from the closest private car and participating taxi at the same price. So there’s private cars with independent drivers not affiliated with a taxi company as well as taxis from participating fleets which included about half of taxis in Singapore at that time.
“Now, remember that at the time of the study, not all taxi companies allowed their taxis to use Grab services. So the supply of taxis in our study is based on operators that forbid their taxis from participating in the JustGrab service.
“Unfortunately, Grab doesn’t want to share with us their data. We understand. It’s proprietary. So what my professor did was to use a proxy for JustGrab supply.”
“What is a ‘proxy’?” asked Teng, scratching his head.
“Ahh . . . A ‘proxy’ means a ‘substitute’. In our study, because we could not get data from Grab, we used taxis from taxi operators that permit their taxis to participate as JustGrab cars as a measure of the supply of Grab cars.”
“Is that accurate?” asked Teng. “JustGrab has other drivers too.”
“You are right. Our proxy for Grab supply is not perfect because private drivers are not included. Also, taxi drivers from participating Grab operators can choose whether to drive as a Grab driver or not. However, since at the time of the study, participating operators do not have other ways of booking rides through a smartphone app, the affiliated taxi drivers are most likely to choose to drive as a Grab driver,” explained the young man.
“So even though we do not have data from Grab on direct rides, we can study the relationship between Grab and taxi bookings due to the overlap of Grab cars and certain taxi operators in Singapore.
“Moreover, since the taxis that operate using the Grab platform are able to observe the surge factor, their driving behaviour should be similar with those of private car drivers. But then like you said, we do not have all the data. Not having the private car data from Grab means that our Grab supply is likely to be underestimated. It’s not perfect. Ideally of course my professor would love to have Grab data. We wish Grab would share with us some of their data. As academics, my professor will rigorously analyse the data and help Grab and the society with the findings.
“Anyway, I won’t go into the technical mumbo jumbo. The bottomline is that statistically, we can control for this bias in our analyses.”
Teng wasn’t so sure that he could intelligently follow how the data were analysed. But he was eager to learn the findings. After all, he didn’t really care about technical details. He’d leave that to the experts. All he wanted was to know the results and how they affected him as a taxi driver, and his family and friends as commuters.
“OK, are you ready?” asked the young man as he wanted so much to share the findings. “We found that Singaporeans are not very sensitive to
