E.1Can the Business Pay for Its Own Succession?

Yes, for a stall with long queues. For a typical stall, the first draft of the Succession Mortgage and the Family Name Royalty would together have taken more than half of the successor's income. A cap on the payment and a protected income for the successor fix that without changing the idea behind it. The plan now has both.

E.1.1One stall, three sets of terms

In the first draft the successor paid the master for the stall's goodwill – its name and regular customers – at 12–18 months of the master's verified profit (checked by an accountant). He paid it as 8% of takings each month, never less than $900 or more than $2,200, until it was paid off; on top of that, the master's family received a royalty of 2.5% of takings from the start (E.4, E.5). The model tests those original terms at 15 months of profit on a tendered NEA stall (won at NEA's open tender, usually by the highest bid) at typical sales (156 plates a day) and at a busy stall with a queue (220 plates a day).

E.1 Can the Business Pay for Its Own Succession?: table 1
Case, per month	Sales	Seller's net	Goodwill	Payment	Years	Successor after payment	After royalty too
Typical stall, original terms	$17,667	$3,169	$47,536	$1,413	2.8	$1,756	$1,314
Stall with a queue, original terms	$24,915	$7,009	$105,131	$1,993	4.4	$5,016	$4,393
Typical stall, original terms, with the package	$17,667	$3,169	$47,536	$1,502	2.6	$3,210	$2,741
Typical stall, original terms, successor keeps 90% of custom	$17,667	$3,169	$47,536	$1,272	3.1	$961	$564
Typical stall, payment capped at 30% of net (adopted)	$17,667	$3,169	$47,536	$951	4.2	$2,218	$1,777
Typical stall, capped, royalty deferred (adopted)	$17,667	$3,169	$47,536	$951	4.2	$2,218	$2,218
Capped, royalty deferred, successor keeps 90% of custom	$17,667	$3,169	$47,536	$670	5.9	$1,563	$1,563
Capped, royalty deferred, with the package	$17,667	$3,169	$47,536	$1,414	2.8	$3,298	$3,298
Subsidised stall, original terms, successor from outside the family (rent steps up to the $1,250 market rent)	$16,232	$3,369	$50,538	$1,299	3.2	$1,751	$1,345
Subsidised stall, original terms, family successor (keeps the subsidised rent)	$16,232	$3,369	$50,538	$1,299	3.2	$2,071	$1,665

Red cells: the successor would earn less than the $2,220 Progressive Wage (the pay ladder the Government sets for lower-wage jobs) for an assistant, before MediSave (the part of CPF that pays for health care) and tax. Goodwill at 15 months of the seller's verified net (12–18 months gives 2.2–3.4 years at typical takings and 3.5–5.3 with a queue). The payment tracks the successor's own takings. A successor from outside the family pays the market rent NEA (National Environment Agency, which runs the public hawker centres) assesses (NEA's HSS FAQs, 14 Mar 2022). When a subsidised stall is assigned, NEA phases that rent in over three years (NEA's assignment form, 11 Dec 2020), and the plan asks NEA to confirm the same for HSS (Hawkers Succession Scheme, NEA's scheme for passing a stall to a non-family successor) successors (E.3); the row shows year one of that phasing, up to $1,250, the median (the middle figure: half are above it, half below) of NEA's assessed market rents. Immediate family keep the pioneer's subsidised rent: it "can be transferred to the immediate family members of the pioneer hawkers at the same low rent" (Dr Koh Poh Koon, 13 Nov 2024). The goodwill sum is the same in both rows, because it rests on the seller's profit, not the successor's rent; the family successor gains only room to pay, $320 a month in the first year.

Why the royalty is on takings. Nigel Chua followed six hawker brands as they expanded (IPS, June 2024, pp. 31–48). Most of the licensed or acquired expansions shrank or closed, and the only licence that paid the founders a share of profit, Lao Chaozhou's, paid them nothing before its outlet closed in November 2023. So the royalty here is a share of the successor's takings at the stall the master built, and the test counts no licence income for the master's family. Where the plan does count licence income, from the Heritage Plate, it assumes that a third of licensees fail and pay nothing in any year (ASSUMPTION, the Budget Book, B.9.5).

E.1.2Four thousand hand-overs

One stall cannot show the risk. So the test takes 4,000 tendered NEA stalls from the model's simulated stalls and keeps only those whose master makes at least $1,500 a month (the middle one makes $3,437). It prices the goodwill at 12–18 months of that profit (the middle price is $50,776) and follows each successor month by month for up to ten years. The successor keeps 75–105% of the master's customers (mean 92%). Prices, wages, utilities and demand move year to year, and rent renews every three years. A successor quits when take-home pay averages under $1,000 over six months, or for a reason unrelated to money (2% a year). Every rule meets exactly the same futures, and every hand-over is also run with no payment at all. That control matters: a successor who would have quit anyway is not the payment's fault. The burden we are looking for is the extra quitting the payment causes.

FIGURE E.1 · SUCCESSORS WHO QUIT WITHIN FIVE YEARS UNDER EACH PAYMENT RULE · 4,000 SIMULATED HAND-OVERS, SAME FUTURES FOR EVERY RULE

E.1 Can the Business Pay for Its Own Succession?: table 3
Rule	Quit in 5 yrs	Due to payment	Profit paid (worst 10%)	Payoff yrs (slowest 10%)	Year-1 pay, worst 10%	Retiree recovers	First-loss cost
No payment (control)	34%	–	–	–	$1,129	–	–
First-draft terms	74%	40%	67% (130%)	3.8 (4.8)	−$344	91%	$19,500
Capped terms	49%	16%	31% (54%)	4.7 (6.2)	$536	84%	$7,500
Protected income $1,000	47%	14%	30% (32%)	4.7 (6.2)	$954	83%	$7,100
Protected income $1,500 (adopted)	34%	0.4%	28% (31%)	4.9 (7.2)	$1,117	85%	$5,900
Protected income $2,000	34%	0.0%	21% (30%)	5.1 (7.9)	$1,129	78%	$6,400

Profit paid: median share of the successor's monthly profit taken by the payment, royalty and the 1.5% escrow (a neutral account that collects the payments and pays them out) fee while the mortgage runs, with the worst 10% of hand-overs in brackets. Payoff: median years, slowest 10% in brackets. Retiree recovers: payments received plus any first-loss payout, as a share of goodwill, averaged over every hand-over. First-loss cost: the guarantee's expected payout per mortgage, up to 24 payments. Pay is before MediSave and tax.

The test answers the question directly. The first-draft terms were a heavy burden: they caused 40% of successors to quit within five years who would otherwise have stayed. The capped terms (payments limited to 30% of profit, with a $600 minimum, but no protected income) still cause 16% of successors to quit who would otherwise have stayed. Two parts of the rule do it. The $600 minimum takes a fixed sum from a successor whose profit has fallen. And the payment rests on a twelve-month average of profit, which is slow to notice a sudden drop, so the payment stays high just when the successor can least afford it. That is the unnecessary part of the burden: it harms the successor and, through the failures it causes, the retiree and the guarantee as well.

E.1.3Through 2020–26

Singapore publishes no stall-level record of takings or hand-overs, so a true backtest is impossible. The substitute is a replay: a typical tendered stall handed over in January, priced on a normal year's profit, then run through what actually happened – the circuit breaker and dine-in bans, the cost surge of 2022–23 (hawker-food prices up 5.7% and 6.1%), and the easing since. The successor keeps 95% of the master's custom.

E.1 Can the Business Pay for Its Own Succession?: table 4
Hand-over	No payment	First-draft terms	Capped terms	With the protected income
January 2020, rent waivers only	quits in month 17 · $1,371	quits in month 6 · $184	quits in month 6 · $798	quits in month 14 · $1,218
January 2020, with income relief	stays · $2,121	quits in month 6 · $684	quits in month 15 · $1,475	stays; pays off in 7.1 years · $1,661
January 2022	stays · $1,767	quits in month 6 · $63	quits in month 6 · $926	stays; pays off in 5.2 years · $1,664

Each cell: what happens to the successor, then average take-home pay in the first year. NEA's relief is included: eight months of rent waiver and six months of table-cleaning fee subsidies from April 2020 (MSE, 3 Aug 2021). The second row adds the Self-Employed Person Income Relief Scheme, $1,000 a month for nine months of 2020. Neither row includes the one-off $500 grants of 2020 and 2021. Demand during COVID, the months of relief and the 2022–23 ingredient surge are assumptions shaped on the published timeline.

A hand-over in January 2020 met the circuit breaker in its fourth month. With only NEA's relief, no rule gets the successor through. With no payment the successor quits in month 17; under the capped terms, quits in month 6; under the protected income, quits in month 14. The last quits sooner than with no payment only because of the payments made from January to March, before the shock. With the income relief scheme as well, the control stays. Under the capped terms the successor quits in month 15; under the protected income the successor stays and pays off the goodwill in 7.1 years. With the relief Singapore actually gave, the protected income gets the hand-over through the pandemic and the capped terms do not. The 2022 hand-over is the sharper lesson. Priced on a normal year and then hit by the cost surge, the successor earns about $1,767 a month in the first year with no payment and stays. Under the capped terms the floor pushes take-home pay below $1,000 and the successor quits in month 6. With the protected income the successor stays and pays off the goodwill in 5.2 years.

E.1.4The protected income

What drives the burden. We then asked which factors push up the share of profit that goes on the payment under the capped terms, with a regression across the 4,000 hand-overs (each factor measured in a typical-sized step, one standard deviation). A thinner profit margin adds 12 percentage points to the share for each step, and losing customers adds 10 points. The size of the goodwill barely matters (+0.1 points), because the payment rate sets the monthly burden and the goodwill sets only how long it lasts. The burden falls hardest, in other words, on the successors least able to carry it: thin-margin stalls, and cooks who have not yet won the customers back. Under the protected income, the extra quitting no longer depends on any of these factors (R² 0.00, where 0 means no link at all).

Choosing the level. A protected income of $1,000 barely helps (14% of successors still quit who would not have), because it sits at the point where a cook gives up. At $1,500 the extra quitting all but vanishes (0.4%). At $2,000 it is gone, but 8% of mortgages are still unpaid after ten years and the retiree recovers 78% of goodwill on average, against 85% at $1,500. The result holds when the assumptions move: with the quitting point at $800 or a successor who keeps 85% or 100% of the custom, the capped terms add 15% to 19% extra quits and the protected income under 1 point. One assumption can break it: if cooks give up at $1,500 rather than $1,000, a $1,500 protection adds nothing (20% extra quits). The protected income must sit above the pay at which a cook walks away, so the escrow should review it each year against what successors actually report.

Paperwork. The escrow already needs monthly takings for the 8% limb and a twelve-month verified profit for the 30% limb. The protection adds one figure, the last three months' profit, from the same records. There is no new data feed.

The terms now in the plan (E.4, E.5)

Payment: each month the successor pays the lower of 8% of takings and 30% of verified profit, but never less than $600 or more than $2,200 – unless the next rule applies.
Protected income: the payment can never leave the successor with less than $1,500 of profit a month, tested against the lower of the last three months' and the last twelve months' profit. In a lean quarter the payment shrinks or stops, the $600 minimum is set aside, and the loan runs longer; anything still owed after ten years is cancelled.
Royalty: starts only once the mortgage is paid off, can never leave the successor below the same $1,500, and runs for ten years at most; the successor may buy it out at any time at a published multiple.

Deferring the royalty costs the master's family nothing in the test: it starts only after the mortgage is paid off, and the test follows each hand-over for just ten years, so a ten-year royalty never runs its full course inside it. Against the capped terms, the protected income cuts the quitting the payment causes from 16% to 0.4% of successors, lengthens the median payoff from 4.7 to 4.9 years, raises what the retiree recovers on average from 84% to 85%, and lowers the guarantee's expected cost from $7,500 to $5,900 a mortgage ($3,200 for a stall like the E.4 worked example). Every party is better off except the retiree's timing, which slips a few months. Across the 4,000 hand-overs the middle half pay off in 4.2 to 5.9 years, which is where the plan's "typically four to six years" now comes from.

What the test cannot prove

The quitting point and the share of custom a successor keeps are assumptions, and they set the level of every failure figure here. With no payment, 34% of simulated successors quit within five years. For comparison, of the 99 incubation stalls NEA had allocated to new hawkers by June 2025, 37 had become full hawkers. A successor taking over a going concern should do better than a newcomer; if so, every failure and cost figure here falls, and the comparison between rules holds. The first draft assumed about 15% of mortgages would fail. In the test, a fifth of successors at strong stalls like the worked example, and two-fifths across all tendered stalls with goodwill to sell, leave before the goodwill is paid, most for reasons the payment does not cause. Each failure costs the guarantee less than its maximum, because a failed mortgage has usually been part-paid: $14,900 on average against a maximum of $30,240 in the worked example.

E.1.5How big the guarantee must be

The Plan's Part VII budgets $660,000 over Years 1–3 for the escrow and its first-loss reserve ($150,000, $240,000 and $270,000). At about $5,900 of expected first-loss cost a mortgage, that looks like enough for about 111 mortgages. That is the wrong test, for four reasons.

An expected cost is an average; a reserve must cover the bad tail. A book of 63 mortgages costs $374,000 on average, so in about half of all futures it costs more. To be 95% sure of covering it, the reserve needs $482,000.
Part of the $660,000 is not reserve. It also pays to set up and run the escrow. If that takes $100,000 (an assumption: $40,000 to set up, $30,000 a year to run), $560,000 is left for losses.
Losses arrive over several years. Of all first-loss payouts, 33% fall in the first year after signing, 20% in the second, 15% in the third, 13% in the fourth and 19% later, long after the Year 3 budget ends.
Failures bunch in a crisis. The 2020 replay, on NEA's relief alone, fails under every rule (E.1.3). If a shock like it struck, every mortgage in its first two years could fail at once, and the guarantee would pay up to 24 payments on each: about $22,000 a mortgage, against $5,900 expected.

E.1 Can the Business Pay for Its Own Succession?: table 7
Mortgages in the book	Expected payout	P90	P95	P99	Crisis: all in first two years fail
5	$30,000	$56,000	$61,000	$78,000	$110,000
10	$60,000	$93,000	$104,000	$126,000	$220,000
20	$118,000	$165,000	$180,000	$209,000	$440,000
40	$238,000	$305,000	$324,000	$364,000	$879,000
63	$374,000	$458,000	$482,000	$530,000	$1,384,000
100	$593,000	$698,000	$729,000	$790,000	$2,198,000

Total first-loss payout over each mortgage's life, adopted terms, 20,000 books drawn from the 4,000 simulated hand-overs. Crisis: every mortgage fails in its first two years and the guarantee pays 24 of its own scheduled payments (at least $600), capped at its goodwill. Red: above the $560,000 left after set-up and administration. Payouts are assumed to fall when a mortgage fails.

A reserve that is topped up, not spent once

The first-loss reserve should be one lump sum that never expires, not a yearly budget line: what is not paid out stays, and what is paid out is replaced. Keep it at or above the amount that would cover the open mortgages' losses in 95 futures out of 100 (the 'P95' payout) and, where affordable, at or above the crisis payout for mortgages still in their first two years. Top it up from operating surpluses and the escrow's 1.5% fee (only about $160 a mortgage a year). The Years 1–3 money can safely cover about 74 mortgages at P95 (89 if all $660,000 were reserve), but only about 25 through a 2020-style crisis (30). For the pilot that is enough. The gate needs six hand-overs across the pilot centres, and a realistic pilot has 5–20 mortgages: 20 mortgages need $180,000 at P95, well inside the reserve, and its first 25 or so mortgages are covered even in a crisis. Growing past about 74 mortgages needs more capital: about $7,300 a mortgage at P95 for 100 mortgages.