24owlsGroup Pte Ltd
← Back to journal
Blog

What Did Singapore's PSG Grant Actually Buy?

About 36,400 firms took Singapore's PSG grant at S$6,600 each, an implied S$240 million. Here is what that money bought, from whom, and what nobody published.

What Did Singapore's PSG Grant Actually Buy?

What did Singapore's Productivity Solutions Grant actually buy? In short: about 36,400 firms got help over three years, at roughly S$6,600 each, and nearly 95% of complete applications were approved. Multiply the two numbers the government did put on record and you get an implied S$240 million spent, a total no ministry has ever published. This guide explains, in plain terms, where that money went, why the same software could cost 25 times more from one vendor than another, and the simple numbers that would let anyone judge whether it was worth it.

The grant worked like a shopping voucher. The government pre-approved a list of software and equipment, set fixed prices, and paid part of the bill almost automatically. That design made it easy to use, but it also means the price list, not any spending report, is the clearest record of what happened.

How much money moved? An implied S$240 million never published

The size of PSG has to be worked out, because no one disclosed it. "In the last three years, about 36,400 companies have received support under the Productivity Solutions Grant (PSG). The average quantum of support for PSG was about S$6,600 per project. Close to 95% of all complete submissions were approved," then trade minister Gan Kim Yong told Parliament in February 2024 [1].

Multiply those and you get an implied S$239.8 million. Keep in mind that is a floor, not an official total, since some firms took more than one project. The last programme-level budget ever published was the S$110 million launch amount in 2018 [2]. Since then, only application counts have surfaced, such as over 116,100 applications across three years [3].

With approval almost automatic and cheques small, value for money was really decided by one thing: what the pre-approved vendors charged.

Why the same software costs up to 25 times more?

Similar tools on the list can differ in price by a huge margin, and nothing on the public record explains why. One HR software listing runs from S$5,700 to S$32,000 [4]; another vendor's packages run S$19,430 to S$37,205 [5]. The government pays half of the approved cost, up to S$30,000 a year [6].

Look across a whole category and the gap widens. One comparison site counts HR systems from S$1,500 to about S$37,000, and accounting tools from S$430 to S$54,140 [9]. When an MP asked what pushes vendors to lower prices as technology gets cheaper, the answer named no mechanism, only that "variety promotes price and quality competitiveness" [10].

Do 576 listings mean 576 products? Not really

The directory looked large: 576 live listings when checked on 17 August 2026 [7]. But many are the same underlying software sold by different resellers. More than 20 listings are versions of the same accounting tool, priced from S$611 to S$10,360 [9]. That is a rough reading from a comparison site, not an official count, but the point stands: the list looks more varied than it is.

Which vendors actually captured the money is impossible to tell from public records. The directory shows vendors and prices but publishes no figures on how much each was paid, and counts even disagree on the basics: one tracker logged 593 solutions [12], another about 549 [11]. The government's open procurement dataset covers its own buying, not these grant-funded purchases [13]. So no outsider can compute where the S$240 million really went.

Who the S$6,600 voucher suited? Small firms buying everyday tools

The money went mostly to very small firms buying standard back-office tools. Official figures, frozen at June 2020, show micro firms were 55% of users and services businesses over 80%, with sales, inventory and customer-management tools each taking about 20% of adoptions [14]. A S$6,600 average is a small-firm-sized cheque, which frames what it could realistically buy.

More recently, almost 3,000 firms adopted AI-enabled tools in 2024 [15]. Yet the very smallest firms often could not qualify at all. They make up 94% of Singapore enterprises but fail the grant's thresholds, which is why a separate S$10 million pool and bank financing were later set up for them [16].

How much was checked? About S$8.7 million of roughly S$240 million

Only a sliver of spending was audited, and the total clawed back was never revealed. "More than 1,100 approved PSG projects, with a total grant quantum of S$8.7 million, were subjected to post-disbursement audits... About 3% of the projects... were assessed to be non-compliant," minister Gan Kim Yong wrote in February 2025 [18].

Put in context, S$8.7 million is under 4% of the implied S$240 million, and the 1,100 audited projects are about 1% of applications. So the reassuring "3% non-compliant" figure describes only that small sample, not the whole scheme. When an MP asked how much money was actually recovered, the reply gave process language and no number [18]. A later reply confirmed vendors can be removed or barred, but disclosed zero counts of any being removed [19].

Did it raise productivity? The only study stops at 2020

The one proper evaluation of PSG ends at 2020 data. It found "an increase in firms' productivity of 3.0 per cent" from PSG, with industry-specific tools nearly twice as effective as generic ones [20]. That study predates today's much larger directory and the AI wave, and no newer evaluation has been published [2].

It also turns out firms are never asked to report the results. "PSG recipients are not required to report the wage increases arising from the solutions," the ministry noted [21]. What exists instead are surveys of opinion: "85% of participating SMEs reported time savings," MCI told Parliament [22]. That is what firms felt, not what was measured.

What can be said in the grant's favour?

To be fair, the positive case deserves a straight hearing. Smaller firms seem to have gained the most. The OECD, restating the same 2020 study, notes "micro firms achieving up to a 6.4% increase in productivity" [2]. And a senior minister of state said firms "reported cost savings of 48% per solution" between 2018 and 2023 [23].

Two honest caveats. The 6.4% and the 48% both trace back to self-reported figures, not independent measurement [24], and a separate sample of 129 firms found over 80% reported some benefit [25]. So the favourable case rests on older data and on what firms said about themselves. Nothing tests the 2021 to 2026 years.

What replaces PSG? The bigger EDGE grant

From the second half of 2026, a new grant called EDGE takes over and raises the stakes. It merges three older grants into one and lifts the yearly ceiling to S$100,000, more than triple PSG's S$30,000 [26]. It also opens up to larger companies, not just SMEs [27].

AI is being pushed to the front. The share of AI-enabled tools on the list is set to rise from 30% to 50% [29], with a pledge to add "safeguards... whilst trying, at the same time, to not make the rules too onerous" [30]. The same pre-approved-vendor machine now runs at triple the size, which is exactly why the old pricing questions matter more, not less.

There is a quieter lesson for business owners here. A grant can buy a tool, but it ends, and whatever it was meant to kick-start has to survive without it. That is just as true of marketing: firms that build their own search visibility and paid-media discipline keep compounding after the subsidy stops, while those that treat it as a one-off purchase restart from zero. Closing that gap is the kind of work an agency like 24owlsGroup, a Singapore digital marketing and branding agency running SEO, Google Ads, social media and event management for SMEs in Singapore and Malaysia, exists to do.

The five numbers that would settle it

The scheme pays out after just one month of proven use [6], under listings that state they are "not to be taken as a form of endorsement" [33]. Five simple disclosures would let anyone judge value for money: how much each vendor was paid, the total clawed back, an evaluation using data after 2020, saved snapshots of the directory over time, and counts of vendors removed or barred. None is published.

Until one of them is, the vendor price list stays the only complete record of what S$240 million of subsidy bought, and EDGE now re-runs the same experiment at triple the size.

Frequently asked questions

How much did Singapore spend on the PSG grant?

No official total was ever published. Multiplying the two figures the government gave, 36,400 firms at about S$6,600 each, gives an implied S$240 million over three years [1]. That is a floor, since some firms took more than one project.

How much does the PSG grant pay?

It covered half of an approved solution's cost, up to S$30,000 per company per year [6]. From late 2026 the replacement EDGE grant lifts that ceiling to S$100,000 [26].

Why do similar tools cost so much more from some vendors?

No public benchmark explains it. Listings for similar software range from about S$1,500 to S$37,000, and the government named no mechanism to keep prices in line, citing only vendor "variety" [9][10].

Did the PSG grant actually improve productivity?

The only proper study, using 2017 to 2020 data, found a 3.0% productivity gain, up to 6.4% for micro firms [20][2]. There is no evaluation after 2020, and firms are not required to report results [21].

Who benefited most from PSG?

Mostly small and micro firms buying standard back-office tools; micro firms were 55% of users and services businesses over 80% [14]. The very smallest firms often could not qualify, prompting a separate support pool later [16].

What is the EDGE grant replacing PSG?

EDGE merges three grants into one shopfront, raises the yearly ceiling to S$100,000, and opens to non-SMEs, from the second half of 2026 [26][27]. It also increases the share of AI tools on the approved list from 30% to 50% [29].

References

Got the next one?

Tell us about your brand, campaign or event. We'll bring the team.

Start a project →