Finance · AI
HONG KONG, Oct 28, 2026 - Hong Kong's Transport Department has a number that nobody in the argument disputes. Its survey of personalised point-to-point demand puts the territory at about 880,000 passenger journeys a day. Taxis carry roughly 690,000 of them, or 78 percent. Ride-hailing carries about 190,000, or 22 percent. Those three figures are the closest thing this debate has to a shared fact, and every party has been quoting them for eighteen months.
The disagreement starts one step later, at the conversion. The government has decided that 190,000 daily ride-hailing journeys should be served by no more than 10,000 vehicle permits. The platform most affected says that cap cuts its active driver pool by more than two-thirds. The taxi trade says 10,000 is too many, too fast, and should be released in batches. Three parties, one demand figure, three incompatible conclusions — and, unusually for a regulatory fight, a paper trail detailed enough to check the arithmetic.
What each side committed to, in its own words
Most coverage of this file has quoted the bill's passage and the fee schedule. The positions are more useful than the statutes, because each one contains a falsifiable commitment.
Speaker
Position
What they said
What it commits them to
Mable Chan, Secretary for Transport and Logistics
Government, selling the bill
The bill would "modernise our services, attract younger drivers, and provide diverse traveling options for residents"
That supply expands, not just legalises
Transport and Logistics Bureau spokesman
Government, defending the cap
10,000 permits is "a prudent, sound and appropriate approach that balances considerations regarding road resources and the public transport ecosystem"
That road capacity, not demand, is the binding constraint
Andrew Byrne, Uber global head of public policy
Platform, opposing the cap
A quota in this range is "very unusual" by the standards of London, Australia and mainland Chinese cities
That Hong Kong is an outlier, not a normal case
Nicole Lee, Uber Hong Kong public policy head
Platform, on cost
"70% say the proposed insurance costs would push them off ride-hailing"
That the insurance rule, not the cap, may bind first
Taxi Owners' Association
Trade, opposing the cap
Government "must absolutely not release all 10,000 permits at once within a short period"
That phasing protects incumbent income
Hong Kong commentator, Chinese Association of Hong Kong and Macao Studies
Pro-establishment, against the licence class
Government should issue more taxi licences rather than create a parallel ride-hailing permit
That the two services are substitutes, not complements
Read the third and fifth rows together and the shape of the fight appears. Uber's argument is comparative: no other dense city caps ride-hailing this tightly. The taxi trade's argument is distributive: whatever the number, it should arrive slowly enough that existing drivers absorb it. Both can be true. Neither addresses the government's actual claim, which is about road space.
The arithmetic the government published
Start with the vehicles. Hong Kong has 18,163 taxis — 15,250 urban, 2,838 New Territories, 75 Lantau — carrying close to a million passengers a day. Add the 30,000-plus drivers Uber says are active on its platform, and the current point-to-point fleet is roughly 48,000 vehicles serving 880,000 journeys. That is 18.3 journeys per vehicle per day across the whole system.
Now impose the cap. The fleet becomes 18,163 taxis plus 10,000 permitted ride-hailing cars, or 28,163 vehicles, for the same 880,000 journeys. That is 31.2 journeys per vehicle per day — a 71 percent increase in vehicle productivity, required immediately.
Scenario
Vehicles
Daily journeys
Journeys per vehicle
Current, taxis plus active ride-hailing
~48,200
880,000
18.3
Cap at 10,000 permits
28,163
880,000
31.2
Cap at 15,000 permits
33,163
880,000
26.5
Thirty-one journeys a day is not fantasy. Hong Kong taxis already achieve about 38 per cab, because street-hailing produces short trips with almost no idle time between fares. The question is whether a pre-booked ride-hailing model can hit the same productivity. It operates differently by design: the car travels to the passenger, waits, and often deadheads back. Independent estimates put deadhead mileage for app-dispatched fleets at 25 to 40 percent of distance travelled. Street-hailing taxis cruise for fares too, but they do it while available, not while committed.
So the cap is not arithmetically impossible. It requires ride-hailing to become as productive as street-hailing. That is precisely what the taxi trade has argued it should be, and precisely what ride-hailing's operating model makes difficult.
The arithmetic the platform published
Uber's numbers point at a different variable: the driver pool, not the vehicle count.
The company says it has more than 30,000 active drivers in Hong Kong. A 10,000-permit cap therefore removes more than two-thirds of them, and its submission to LegCo states that over 20,000 current drivers would lose the ability to earn through the platform. Its survey of 4,569 drivers found 74 percent "extremely concerned" about the quota and a further 18.5 percent "very concerned"; 41 percent said ride-hailing provides between 81 and 100 percent of household income; 76 percent expected immediate financial difficulty.
Those are self-reported expectations from a platform's own user base, collected in May 2026, and they should be read as such. A survey commissioned by the party that loses most from a policy will find that the policy is unpopular. The useful part is not the sentiment, it is the structural claim underneath: that the permit is attached to the vehicle and bound to its owner, so the number of permits is a hard ceiling on the number of people who can earn.
That claim is correct on the text. The owner-driver requirement means a permit holder must be the registered owner of the car they drive, a provision Mable Chan said was intended to stop permits being sublet or sold on a secondary market. It is a strong anti-rent-seeking design. It also means there is no legal way for a fleet owner to hold twenty cars and employ twenty drivers, which is how ride-hailing scales in most markets.
Uber's proposed fix is a phased allocation with 80 percent of permits reserved for existing active drivers and no more than 20 percent for people who already own a car, plus quarterly reviews using 90th-percentile indicators on wait times, completion rates and fares. The trade's fix is the opposite: batch release, and a floor under ride-hailing fares so they cannot undercut taxis.
The arithmetic the trade published
The Taxi Owners' Association's submission is the shortest and the most revealing. It asks for two things: that the 10,000 permits be released gradually rather than at once, and that ride-hailing fares be prohibited from going below taxi fares.
The second request is the important one, and it is not really about fares. It is an admission about the first. If the trade believed 10,000 permits were a small number, it would not need a price floor. A price floor only matters when you expect the new entrant to win on price at equal volume. The association is asking the government to remove the one advantage ride-hailing has, which suggests it expects the volume to be there.
The first request has a defensible economics behind it. Taxi plates in Hong Kong are traded assets. Anything that reduces expected earnings on a plate reduces its capital value, and a sudden release of 10,000 competing vehicles would do that in a single step. Phasing spreads the loss over time and lets incumbents exit through attrition rather than writedown. That is a legitimate transition-management argument, and the government has effectively accepted it by setting the permit regime to start in August 2027 rather than with the bill's commencement.
What the trade did not supply is any figure for how many permits would be acceptable. That absence is itself a position.
What the government got for the taxi trade, quietly
The bill was not only a cap on ride-hailing. It also carried three concessions to the taxi industry that were legislated separately and are already in force, and they change the competitive picture more than the permit number will.
Since 1 April 2026, taxi drivers must offer at least two electronic payment methods, one of them QR-based and one not. Five licensed taxi fleets, awarded under the fleet scheme the Transport Department introduced to raise service standards, are now operating with app-based booking. The combined driving test for taxi and ride-hailing vehicles, introduced in the same subsidiary legislation package, lets a single qualification serve both.
Measure
Status
Effect on the 2027 market
Two e-payment methods per taxi
In force 1 April 2026
Removes the "no card, no app" objection to taxis
Five licensed taxi fleets
Operating
Gives taxis app dispatch without a platform licence
Combined taxi and ride-hailing test
In subsidiary legislation
Makes driver supply fungible between the two
In-vehicle cameras
Rolling out
Raises the compliance floor for both modes
That third row is the one to watch. If a driver can qualify once and choose between a taxi and a permitted ride-hailing car, then the 10,000 permits are not competing against 18,163 taxis for labour. They are competing inside a single 28,000-vehicle pool for the same drivers, on the same test, with the taxi option carrying no permit queue.
Sixteen thousand applicants, ten thousand permits
There is a number in this file that has received almost no attention, and it is the one that predicts what happens next.
More than 16,000 people have already applied to sit the ride-hailing driver test, according to reporting from the mainland financial press on the eve of the licensing round. The permit ceiling is 10,000. Even if every permit is issued on day one — which the batch-release argument is designed to prevent — at least 6,000 people who have paid to qualify will have no vehicle to drive.
That surplus has three possible destinations. It can queue for permits released in later tranches. It can move into taxi driving, which the combined taxi and ride-hailing driving test makes easier by design. Or it can exit, having sunk the cost of a course and a test into a licence class that turned out to be rationed.
The government has been explicit that it is building the queue on purpose. Its requirements for platform applicants — a track record in a city of 7.5 million or more, an average of 100,000 daily orders across two cities, HK$50 million paid-up capital, HK$30 million in bank deposits and HK$30 million in credit facilities, three years of audited accounts, a permanent office and staff in Hong Kong — are not tests of whether a platform can operate. They are tests of whether it can absorb a bad year. The HK$1.2 million annual licence fee is trivial against those balance-sheet requirements. The application is a filter, and the filter is capital.
Mean capacity is fine. Variance is not.
Here is where the arithmetic gets more interesting than either side's talking points.
At 10,000 permits, the system has enough vehicles to absorb 190,000 daily journeys on average, provided each car works a long shift. What it cannot absorb is the shape of the demand curve. Point-to-point demand in Hong Kong is brutally peaky: the morning build-up, the evening release, Friday and Saturday nights, and above all rain. Uber's rider survey found 86 percent expecting greater difficulty booking in peak hours or on rainy days. That is not a claim about average capacity. It is a claim about the tails.
A fixed permit cap is a tool for managing the mean. It is the wrong tool for the tails, because the tail is served by drivers who are marginal — part-time, second-income, available on the bad night and not on the good one. Uber says 41 percent of its drivers depend on the platform for most of their household income, which implies 59 percent do not. That 59 percent is the flexible layer that covers the rain. A permit system that requires an owner-driver to buy a car and hold a permit will select for the 41 percent and eliminate much of the 59.
This is the strongest version of the platform's case, and it is not the version the platform makes. Uber argues volume. The better argument is elasticity: 10,000 full-time professional drivers will deliver a very reliable Tuesday and a very unreliable typhoon.
The counter-argument is equally strong. The 59 percent flexible layer is also the layer with the least insurance, the least vehicle maintenance and the weakest incentive to comply with the owner-driver rule. A regulator that wants enforcement has a reason to prefer 10,000 known drivers over 30,000 partly-known ones, and the vehicle data management system the Transport Department is building — real-time operational data from licensed platforms, used to assess performance and inform future policy reviews — only works if the population is enumerable.
The question none of them answered
Three questions went unanswered in every submission I read, and they decide more than the cap does.
· Insurance. Uber says 70 percent of drivers would leave if required to buy commercial third-party cover costing HK$10,000 to HK$30,000 a year, and that 84 percent prefer platform-provided blanket cover priced per trip. No party has published what the actual premium would be under a permit regime. Until someone does, the insurance rule, not the permit number, may set the real ceiling.
· Vehicle age. The 12-year cap was relaxed from seven after authorities accepted that seven would squeeze supply. Nobody has published the age distribution of the current active fleet. If a large share sits at nine to twelve years, the first renewal cycle will remove cars faster than permits are issued.
· The cross-hire question. The owner-driver rule forbids subletting. It does not address whether a permitted driver may work across multiple platforms. If they can, the 10,000 permits are 10,000 drivers serving the whole market and the cap bites less than expected. If they cannot, each platform's share of 10,000 becomes the number that matters, and the licence round turns into an allocation fight.
The government has promised the permit quota will be revisited through subsidiary legislation. That is where these three questions will actually be settled, and none of them were in the bill.
What the first licence round will show
The application window closes at noon on 30 October 2026. The licences run five years. The permit regime begins in August 2027.
Between those dates, three observable things will tell you whether this framework works, and none of them is the headline permit number.
The first is how many platforms apply. The capital and track-record requirements are calibrated to admit a small number of large operators and exclude everyone else. If two or three applications arrive, the regime has produced an oligopoly by design and the 10,000 permits will be divided by negotiation. If five or more arrive, the fee and capital thresholds were set too low to do the filtering job they were written for.
The second is the first quarterly data release from the vehicle management system. Wait times and booking success rates by district, at peak and off-peak, are the only figures that will show whether 10,000 was right. The government has committed to using exactly these indicators for review. It has not committed to publishing them.
The third is what happens to the 6,000 surplus test applicants. If they convert to taxi drivers, the policy has done what Mable Chan said it would — modernise the trade and attract younger drivers — and the taxi fleet gets the renewal it has needed for a decade. If they exit, the policy has created a queue and called it a market.
The taxi trade's price-floor request and the platform's 80 percent allocation request are both attempts to control the same unknown: who holds the 10,000 permits when the music stops. The government's answer, so far, is that the market will decide within a ceiling it sets. That is a coherent position. It is not yet an evidenced one, and the evidence that would support it — insurance premia, fleet age, cross-platform rights, peak wait times — is all in the government's hands and none of it is public.
Sourcing note
The direct quotations in this article come from published statements and submissions, not from interviews conducted for it. Mable Chan's remarks were made in the Legislative Council on 15 October 2025 and reported by Hong Kong Free Press; the Transport and Logistics Bureau spokesman's "prudent, sound and appropriate" formulation appeared in the government's gazette statement on the subsidiary legislation; Andrew Byrne's "very unusual" characterisation and the 70 percent fare-increase estimate were given at a briefing in May 2026 reported by local outlets; Nicole Lee's insurance comment came from Uber's own newsroom release of its May 2026 survey; the Taxi Owners' Association's position is taken from its written submission to the LegCo bills committee, translated from Chinese.
Survey figures attributed to Uber — 10,009 riders, 4,569 drivers, conducted May 2026 — are the company's own research and are identified as such throughout. The 880,000 / 690,000 / 190,000 demand split is the Transport Department's survey estimate as reported by local media; the department has not published the underlying tables. The 16,000 test applicants figure comes from mainland financial press reporting and has not been confirmed by the Transport Department.
Sources
· Transport Department 🔗 — Ride-hailing Vehicles: licensing regime, fees, applicant eligibility, accessed October 2026
· Transport Department 🔗 — Taxis: fleet size, daily patronage, electronic payment requirement, accessed October 2026
· Hong Kong Free Press 🔗 — "Hong Kong passes bill to regulate ride-hailing platforms under licensing regime", 16 October 2025
· China Daily Hong Kong 🔗 — "Subsidiary bills on regulation of ride-hailing services in HK gazetted", 2026
· China Daily Hong Kong 🔗 — "Hong Kong opens license applications for ride-hailing services", August 2026
· Uber Hong Kong 🔗 — rider and driver survey findings, May 2026
· Hong Kong Business 🔗 — "Uber says 83% expect higher fares under Hong Kong cap", 2026
· HK01 🔗 — submissions by Uber and the Taxi Owners' Association to the LegCo bills committee, 2026（Chinese-language original）
· Jiemian News 🔗 — "Hong Kong ride-hailing is coming: 16,000 have already registered for the test", 2026（Chinese-language original）
This article was prepared by the expert team at Global Infinity Labs, drawing on insights and data provided through its MCP data services.
About Global Infinity Labs
Global Infinity Labs is an AI innovation company specializing in MCP data services.
Email: support@globalinfinitylabs.com
For Media Inquiry
hk@asiapresswire.com
https://www.asiapresswire.com
Oct 28 · 04:30·ACN Newswire