Skip to content
Silicon Wok Hawker Heritage
Menu
Contents

D.3One Plate, Six Landlords

The same plate of food splits very differently depending on who owns the floor it is cooked on. The model represents the trade as six kinds of stall, one for each kind of landlord the plan identifies (the Plan, §2.1, Hawkers and Social Protection, C.7).

D.3 One Plate, Six Landlords: table 1
StallStallsPlatePlates/dayRent/moFees/moSales/moHawker netNet share
NEA centre, subsidised2,070$4.30150$290$720$16,232$3,36921%
NEA centre, tendered4,330$4.50156$1,250$720$17,667$3,16918%
Operator-run centre500$4.40160$1,700$1,050$17,315$1,5579%
HDB-rented coffeeshop2,800$5.20160$3,400$1,000$20,939$2,47212%
Private coffeeshop3,000$5.50195$6,000$1,200$26,991$1,9997%
Mall food court1,500$6.5015025% of sales$300$24,538$2,74411%

Calibration. The tendered NEA stall reproduces NEA's 2022 cost survey: ingredients 54%, manpower 20%, rent 9% and utilities 7% of operating costs (NEA: 56%, 20%, under 10%, about 7%). The tendered stall pays the $1,250 median rent across NEA's centres; fees are 2025 medians from MSE parliamentary replies. Coffeeshop rents are assumptions: $3,400 at an HDB-rented coffeeshop, against a commercial guide's $4,000–10,000 (Home & Decor, 2026), so the base is conservative. Manpower of $2,920 a month covers one assistant at the July 2026 PWM benchmark of $2,220 with 17% employer CPF (Central Provident Fund, Singapore's compulsory savings for retirement, health and housing), plus part-time help. Delivery orders pay GrabFood's published 30% service fee, before GST. The stall opens 26 days a month, less 10 unwanted closure days a year, at 13 hours a day. Stall counts by type are estimates. We estimate about 6,900 cooked-food stalls in public centres. (In November 2024 MSE spoke of "our 7,000 cooked food stalls"; NEA's rent support counts about 15,000 stalls, cooked food and market together.) About 30% of the 6,900 are subsidised stalls, and we split the count that way. No official count of coffeeshop stalls exists. There are 776 coffeeshops in HDB estates: 374 rented from HDB and 402 that HDB sold before 1998 (Enterprise Singapore, MND and HDB, 1 Jul 2024). At about 7.5 cooked-food stalls each (Kimly's audited 7.2 an outlet; brokers' rule of thumb is 8–10), that gives about 5,800 coffeeshop cooked-food stalls, within a range of about 5,600–7,200. SFA (Singapore Food Agency, which licenses food stalls) licenses 13,893 food stalls across coffeeshops, food courts, canteens, private markets and non-NEA hawker centres (end 2025), which caps the total. Food-court turnover rent of 25% is the weakest parameter in the model.

Where each dollar a diner pays ends upCENTS PER DOLLAR OF SALES · ONE STALL, ONE MONTH · FOLLOW-THE-PLATE MODEL, 2026 BASELINENEA centre, subsidised$4.30 plate4518621NEA centre, tendered$4.50 plate45817618Operator-run centre$4.40 plate461151769HDB-rented coffeeshop$5.20 plate451914512Private coffeeshop$5.50 plate4525137Mall food court$6.50 plate45261211IngredientsLandlordCentre servicesAssistantsUtilitiesPlatforms & paymentsOther running costsHawker household
FIGURE D.2 · CENTS OF EACH DOLLAR OF SALES, BY TYPE OF STALL · OTHER RUNNING COSTS ARE LICENCES, INSURANCE, PACKAGING AND EQUIPMENT UPKEEP
Worked example · a $4.50 plate at a tendered NEA stall, and a $5.50 plate at a private coffeeshop
NEA TENDEREDPRIVATE COFFEESHOP
Ingredients (suppliers)$2.00$2.45
Landlord: rent and fees$0.37$1.37
Centre services (cleaning, conservancy)$0.13$0.10
Assistants' wages$0.74$0.73
Utilities$0.25$0.24
Platforms and payments$0.11$0.14
Other running costs$0.08$0.06
Hawker household$0.81$0.41
The diner pays$4.50$5.50

The private-coffeeshop diner pays a dollar more, and almost all of that dollar goes to the landlord. The hawker's family keeps $0.41 of the higher price against $0.81 at the NEA stall. This is the economics behind the plan's rule that rent policy follows the landlord (the Plan, §2.1): where the state sets the rent, it can set it for heritage; where the market sets it, only the person can be supported without the landlord taking a share.

This is a working paper for public discussion. Every number in it is an invitation to improve it. Corrections and better data are welcome.