D.4What the Hawker Earns per Hour

One official figure measures what hawkers earn, and it is low: about $1,470 a month in 2021, the median for full-time hawkers who work on their own (MOM, in the Advisory Committee on Platform Workers' report, 2022, Fig. 1.1; read from a chart). The model gives a fuller estimate with honest error bars: for most stalls, the person who owns the business earns less per hour than the Progressive Wage benchmark for the assistant they hire.

The computer builds 5,000 stalls of each type. Each gets its own number of customers, prices, ingredient costs, wages, utility bills, hours (11–15 a day), days open, share of delivery orders and closure days, drawn from the ranges in Method, Levers and Data, F.2. We also vary how many customers leave when prices rise, and how much of a cost rise the stall passes on. The middle of each range is the figure we found in a source, and busier sites get higher rents. The package results assume each stall takes up every element of it. The table gives medians and percentiles (the level a given share of results stays below) of the simulated stalls; D.4.1 and D.7 use the base-case stall instead, so a tendered NEA hawker shows $9.67 an hour here (simulated median) and $9.69 there (base case).

D.4.1How much rent can a stall bear?

The papers quote three NEA rents, and they measure different things. The median across NEA's centres, about $1,250, is what a non-subsidised cooked-food stallholder typically pays; it has been stable for about a decade (MSE, 18 Sep 2023 and 5 Feb 2025), and the median assessed market rent (NEA's valuation of what a stall would fetch) NEA gave Parliament is the same $1,250. It is the model's typical tendered stall (D.3). The 2025 median at the centres comparable with the operator-run ones, $1,450 (MSE, 8 Sep 2026), is the figure to set beside those centres' $1,700 (D.2.2). The 2023 median winning bid, about $1,800 (the Plan, §2.1), is what newcomers bid for a vacant stall. None of them is "the" rent, so the model treats them as a range. It holds everything else at the typical tendered stall and moves the rent from $800 to $3,000 a month.

Red cells: below the assistant's $11.64 an hour, or a successor below the $1,500 protected income. Successor: what the successor takes home in one ordinary month under the Succession Mortgage terms the plan now uses, assuming the stall keeps 95% of its customers and the verified profit equals that month's profit. It is a shortcut for the month-by-month test in Succession Economics, E.1. The last two columns are the share of 5,000 simulated stalls below the benchmark at each quoted rent, using the same simulated stalls as D.4's table.

Where it stays sound. The stall itself is robust. A typical tendered stall makes a loss only above about $4,420 a month of rent, and at the three quoted rents only 0.7% to 2.3% of simulated stalls lose money. With the package, the typical hawker earns more than the assistant at any rent up to about $2,450, well above the highest quoted figure. The Succession Mortgage holds up to about $1,810: below that, a typical successor keeps more than $1,500 a month after the payment.

Where it breaks. Without the package, the typical hawker earns less than the assistant at any rent above $610, so D.4's finding holds at all three quoted rents; the rent only decides by how much. Each extra $100 a month of rent costs the hawker about $0.31 an hour. The share of stalls below the benchmark rises from 66% at the median rent to 77% at the winning bid, and with the package from 26% to 37%. At the $1,800 bid the successor keeps about $1,506, close to the protected income; above about $1,810 the protection binds, the payment shrinks and the term lengthens, and at about $2,450 the payment stops altogether.

What each lever (a setting the plan can turn up or down) adds. The package has nine parts. Four come first: price freedom, Makan Vouchers, the Relief-Hawker Corps and the Protection Floor (the minimum savings and cover the plan proposes for every hawker). Five more sit beside them: cheaper ingredients, power and delivery through the Buying Circles (D.16), the table-cleaning fee subsidy (D.14) and the named deputy who covers an hour a day (the Plan, §2.7). With only the first four, the typical hawker still earns more than his assistant as long as the rent is no more than about $1,540 a month. The table shows how far each of the other five, added on its own, raises that limit; all five together raise it to about $2,450, or $2,150 without the deputy. The deputy moves it most, because an hour given back raises what every other hour earns.

Savings and the deputy's hour are assumptions the pilot centres test: ingredients 3% and utilities 10% cheaper for circle members, delivery at a 25% group rate instead of GrabFood's 30% list rate, $200 a month of a pilot-centre stall's table-cleaning fee paid by the Government, and 12 of 13 hours worked (D.16, D.14, the Plan, §2.7). The circles' target for delivery is 15%; at that rate the delivery lever would move the break point (the rent above which a typical stall earns less an hour than its assistant) to about $1,760, but no platform has offered it, so the model does not count it.

What a newcomer actually pays. The medians flatter the newcomer's bill, because a winning bid is paid for longer than one tenancy. Take a hawker who wins a new stall at the 2023 median bid of $1,800, and suppose NEA values it at $1,450 (NEA does not publish its valuations, so this is an illustration). Under NEA's renewal rule of November 2024, which applies to new tenancies only, he pays $1,800 a month for three years, $1,625 for the next three and $1,450 only from the seventh year. By then he has paid about $18,900 more than the valuation (derived). Measured against the $1,250 median across all NEA centres instead, the middle three years cost $1,525 a month and the extra paid comes to $29,700. On the typical stall that is $8.01 an hour for the first three years, $8.54 for the next three and $9.08 from the seventh, all below the assistant's $11.64 (modelled). KF Seetoh says the benchmark is reached "only after six years of paying whatever rent a stallholder tendered" (The Online Citizen, 15 Sep 2026). He is right that it takes six years, but wrong about the middle three: in those years the rent is halfway between the bid and the valuation, not the full bid. The rent bands (D.14) replace the bid itself.

D.4.2The Regulars' List: What a Tier Means in Orders

Profitable hawkers live on their regulars. The Regulars' List (the Plan, §3.6) counts them by name: a stall is listed, at Bronze, when 20 of its regulars put their names to it, and it rises to Silver, Gold and Platinum at 50, 100 and 250 names, or at the share of its weekly orders those names place, whichever comes first.

The unit is the order, not the plate. A regular is one person who says he orders from the stall at least weekly; the model takes him to order 1.5 times a week (assumption). A takeaway for a family is one order of several plates. It lifts the stall's plates without adding a regular, so a share of plates would make a stall that sells by the family look less loyal than it is. Orders a week are the stall's plates a week over its plates an order, and that depends on the dish: about 1.2 for drinks and snacks, 1.3 for noodles and rice, and 2.0 for family dishes such as kway chap, zi char and roast meats (assumptions). Dollars an order are the plate's price times plates an order. No one publishes plates or items an order at Singapore's hawker stalls. The nearest published figure is for e-payments: stallholders on Hawkers Go Digital took $45 million in 5.4 million transactions in August 2023 (Enterprise Singapore, 26 Oct 2023), about $8.33 a transaction (derived), or about 1.9 of the model's $4.50 plates. Treat that as a ceiling, not an average: small orders are more often paid in cash, and coffeeshops and food courts are in the count.

Take the typical tendered stall. It sells about 936 plates of noodles and rice a week, which is 720 orders of $5.85 each. If it has 50, 100 or 250 regulars, they account for 10.4%, 20.8% and 52.1% of its orders. We set the Silver, Gold and Platinum shares at the round numbers just below: 10%, 20% and 50%. Version 1.3 set them at 8%, 15% and 40% of plates; the typical stall reaches each grade at about the same number of regulars, and the same orders, as it did then. From Silver upward, the share route helps a small stall and the head-count route helps a stall with long queues. John, cooking about 630 plates of kway chap a week on his own, takes about 315 orders; he needs 20 names to be listed, like everyone, but reaches Silver with 21. Ah Kow, at 1,200 orders, reaches Platinum with 250, though they place only 31% of his orders.

What healthy looks like. No one has measured it for hawkers; the nearest evidence is from restaurants. Repeat customers brought quick-service restaurants 71% of their 2012 sales, and fine dining 51% (National Restaurant Association, 2013). In Toast and Resy's 2026 data, which count orders, up to half of a restaurant's orders come from about 7% of its guests. Granted, a repeat customer is anyone who comes back, and a regular comes weekly, so the model takes half of the quick-service figure, 35% of orders from weekly regulars, as the mark of a durable base (assumption). At the typical tendered stall that is about 168 regulars: more than Gold needs and fewer than Platinum needs, so a base that healthy puts the stall at Gold. The weakest number is the order rate: at one order a week the typical stall needs 252 regulars for that base, at two 126. NEA's patron survey says only that 83% of residents eat at a hawker centre at least weekly (13 Jun 2019), not how often they return to one stall.

Regulars needed to reach each grade, at 1.5 orders a week each (assumptions). A star (*) means the grade is reached on the share of orders before the head-count. "35% base" is the number of regulars who place 35% of the stall's orders, and the grade they hold. Bronze, the listing, is always 20 names. Orders a week = plates a day × days open a month ÷ (52 ÷ 12 weeks in a month) ÷ plates an order. The six archetypes (a typical stall of one type) are landlord types, so each takes the typical dish, noodles and rice; John and Ah Kow cook kway chap, a family dish.

By dish type. Hold the typical tendered stall's plates and price, and change only the dish. A family-dish stall selling the same 936 plates takes 468 orders, so 32 regulars lift it to Silver; a drinks stall takes 780, and reaches each tier by its count.

Assumptions until Year 0. Every number in this section is an assumption until the plan's first count replaces it (the Plan, §6.1). At every stall in the pilot centres that agrees, listening teams will count orders over two weeks and ask the hawker about his regulars. They will record orders a day, plates an order, dollars an order and the share of orders from named regulars, by dish type. Its first use is to set plates an order and the share thresholds for each dish type before the first listing. The counts route, 20, 50, 100 and 250 names, does not move.

A compliment, not a test. The share route uses the orders a week the hawker reports; no share is published for any stall. A stall that stays at Bronze for ten years has a neighbourhood that wants its dish kept, and the list says only that.