Foreign Caregiver Pilot Got Pricier, Not Cheaper

Of the 100 Who Arrived on August 6, 2024, 73 Remain

On August 6, 2024, 100 Filipino domestic caregivers landed at Incheon International Airport. They were the first cohort of a pilot program for foreign domestic caregivers, proposed by the Seoul Metropolitan Government and launched by the Ministry of Employment and Labor. As of July 2026, 73 of them are still in the country. 27 left before completing their contracts. In 2026 alone, 6 went home in March, and one more departed each month from April through June.

Over the same period, the price of the service moved in the opposite direction. The hourly rate, which started at ₩13,940 (~$10.1), rose to ₩16,800 in 2025 and ₩18,900 in 2026 — a 36% increase in two years.

On December 22, 2025, the government’s Foreign Workforce Policy Committee decided not to convert the pilot into a permanent, nationwide program. A system designed to ease the cost of childcare shut down without ever having eased it.

---

The Original Plan Was for 1,200

The Ministry of Employment and Labor’s original plan was to end the pilot in February 2025 and scale the workforce up to 1,200 by the first half of that year, rolling it out nationwide as a full program. That plan stumbled twice.

The first stumble was on the demand side. Almost no interest came from regions outside Seoul. Once the premise of nationwide expansion collapsed, the shift to a full program was postponed. Instead, the government extended the total allowable stay of the workers who had arrived in 2024 to 36 months and simply extended the pilot period itself — keeping the existing workforce in place rather than bringing in new hires.

The second stumble was in the design of the program itself. In September 2025, the Ministry of Justice announced it would stop operating the ‘gasa-sayongin’ (a household-employer contract arrangement)¹ scheme, under which households contracted directly with foreign workers. It cited criticism from women’s groups, labor groups, and the press over the arrangement’s exemption from minimum wage, low uptake among foreign workers, and how local governments had been running the program. Then in December 2025, the Ministry of Employment and Labor confirmed it would not pursue a full program. At the same meeting, the 2026 quota for foreign workers under the Employment Permit System (E-9)² was cut 38%, from 130,000 to 80,000.

The pilot program covering the 73 remaining caregivers runs through August 6, 2027. After that, it can only be extended by an additional 1 year and 10 months, and only if both the caregiver and the employer agree.

There Was Never a Structural Path to Cheaper Care

There was exactly one reason this program drew so much anticipation: the promise of supplying care at a lower cost than domestic caregivers. But that premise never held up, not even at the outset.

South Korea has ratified the International Labour Organization’s anti-discrimination convention³. Under it, foreign and domestic workers must be guaranteed equal wages. Exempting foreign workers from minimum wage was never on the table. If wages are equal, the only remaining variable is brokerage, training, and management costs—and those rise per person as the number of participants shrinks. In a program that started with 100 people and shrank to 73, there was no way for the per-person cost to come down.

The numbers show this plainly. In 2026, the hourly rate for Filipino domestic caregivers, at ₩18,900 (~$13.6), is higher than the ₩16,620 (~$12) government-subsidized childcare rate that Herald Business, a Korean economic daily, used as its comparison benchmark. The “affordable alternative” turned out to be pricier than the public service it was meant to undercut.

The composition of participating households followed the same logic. According to data the Ministry of Employment and Labor released in 2025, 9 out of 10 participating households earned ₩6,000,000 (~$4,300) or more per month. The ₩9,000,000–₩12,000,000 (~$6,500–$8,600) bracket accounted for the largest share at 30.4%, and households earning ₩18,000,000 (~$13,000) or more made up 23.3%. Households under ₩6,000,000 made up just 8.9%. By residence, Gangnam District led at 19.64%, followed by Seocho District at 13.39% and Seongdong District at 11.61%.

Let me be clear about one thing here: the fact that users skewed toward high earners doesn’t mean the program was designed for the wealthy. It’s closer to the result of only those households able to absorb an hourly rate of ₩18,900 out of pocket remaining in the pool. It wasn’t design intent that filtered the users—it was price.

A Closed Program Doesn’t Make Demand Disappear

Let me bring in a public-sector number here. According to data that Assemblymember Cho Eun-hee’s office obtained from the Ministry of Gender Equality and Family, the average wait time for Aidolbom, Korea’s public childcare-matching service, in 2025 was 39.4 days. The trend runs: 19 days in 2021, 27.8 days in 2022, 33 days in 2023, 32.8 days in 2024. The wait hasn’t gotten shorter.

On one side sat a public service you had to wait over a month for. On the other sat a private-sector alternative priced at ₩18,900 (~$13.6) an hour. And that alternative just closed. There’s no official statistic yet on where the families caught in between went.

Let me flag a search signal that came into the brief here. This is a value observed by an internal search-analytics tool on September 5, 2026. The interest trend for the search term “foreign domestic worker” was declining (trend rate -0.608), while its seasonality was classified as spike-type — surging at specific moments. And the search-intent tag was transactional, even though the actual character of the search results page was classified as informational.

There are three things to be careful about when interpreting these figures.

First, search volume counts searches, not people. If one person searches ten times, that registers as ten.

Second, “search intent” here means a classification tag applied to the query. It isn’t the result of asking actual users what they were trying to do.

Third, this figure is a derived value from an internal tool and hasn’t been cross-checked against public search metrics. Until the raw data and calculation method are confirmed, it should be treated only as a hypothesis.

Read within those limits, here’s what it looks like. Interest itself is declining. A spike pattern that rises only when media coverage is heavy looks more like a reaction to news than anything else. But the combination of “a transactional search term paired with an informational results page” reads differently. It suggests that searches trying to find out how to apply or what it costs are still coming in, while the page only serves up explanations of the program’s nature or notices that it has ended.

Let me say it again: this is a hypothesis, not an observation. The moment you use a search curve alone to claim “demand outside the program is this big,” you’ve gone past the evidence. Search volume proves neither adoption rates, nor market size, nor causation.

What This Program Left Behind, Verifiably

Sticking to confirmed facts, three conclusions remain.

First, without a lever to lower prices, a care-supply policy won’t unfold as designed. If you set “affordable labor” as a policy goal while unable to actually lower wages, all that’s left is the question of who absorbs the management costs. This program shifted that burden onto the families using the service — and what remained were only the families who could afford to pay.

Second, a program’s success or failure shows up first in dropout rates, not headcounts. Once 27 out of every 100 workers left their contracts unfinished, the question to ask wasn’t whether to expand the program — it was why people were leaving. High satisfaction among user families and a 27% dropout rate are two faces of the same initiative.

Third, ending the program was not a verdict on demand. The government’s stated grounds for shutting down the main program were insufficient regional demand and controversy over the system itself — not a finding that care demand had declined. Right next door sits a public service where the average wait time, still unchanged, is 39.4 days.

Oswarld’s Lens

This piece doesn’t draw on any firsthand experience of mine. I’ve never met the families who use this service, nor the operators who run it. What I can do instead is walk through how to read the numbers that are already public.

The first thing I looked at wasn’t the price — it was what the price was being compared against. ₩18,900 (~$13.6) isn’t inherently expensive or cheap on its own. What determines the judgment is what you place next to it. Set it beside public childcare programs and it looks expensive. Set it beside a 39-day waiting list and it looks like a bargain. Whenever a pricing debate splits people into opposing camps — whether it’s about policy or a product — it’s usually because each side is comparing it to something different.

The second thing I looked at was the 27% dropout rate. Satisfaction surveys only ask the people who are still there. Anyone who’s already left has dropped out of the sample. That’s likely why high satisfaction among host families and 27 caregivers returning home midway through the program can coexist within the same initiative. With any service, if you look at user satisfaction without also looking at provider retention, you’re only seeing half the picture.

The third thing is how I treat search signals. I’m generally wary of using search trend curves as evidence of demand. A search spike is a trace of lingering interest, not proof that someone is willing to pay. That said, when the nature of the search terms doesn’t match the nature of what the search results actually deliver, it’s worth a second look — it can signal that people looking to do something are only being served explanations. And in this case, that’s as far as the evidence lets me go.

That said, these three points are my own reading of the data, not a settled conclusion. To turn them into a firm judgment, you’d first need to check the raw search data and the actual basis used for the fee comparisons.

Closing

When you come across news about care policy or services, I’d suggest checking one thing first: does the promise to lower prices come with an actual mechanism for lowering them? Wages, subsidies, or scale — at least one of these has to move for the price to actually drop. Any plan that keeps all three fixed while promising lower prices usually just ends up shifting the burden onto the user.

💬 If you’ve hired childcare or housekeeping help, tell me in the comments what you based your choice on. I’m curious whether price, wait times, or trust was the sticking point for you.

📨 Please pass this along to anyone designing care workforce policy or welfare services.

References & Further Reading

Primary sources

• Herald Economy (Herald Business), “Two Years Into the Pilot, 73 Filipino Domestic Care Workers Remain”, July 14, 2026. ··· Most of today’s numbers come from here — the headcount, fee curve, scheduled end date, and household income breakdown are all laid out in one article.
• Hankyung (Korea Economic Daily), “Filipino Domestic Helper Program Scrapped After Just One Year”, December 22, 2025. ··· Confirms that the decision not to move forward with a full-scale program and the 2026 cut to Employment Permit System quotas came out of the same meeting.
• Bravo My Life, “Ministry of Justice Formally Scraps Controversial Foreign Domestic Care Worker Scheme”, September 8, 2025. ··· Contains the Ministry of Justice’s own explanation for discontinuing the gasa-sayongin (household-employee) arrangement.
• The Public, “Government Childcare Support Service Averages a 39-Day Wait — When Does Private-Sector Participation Begin?”, May 2026. ··· Shows the five-year trend in public childcare waitlist times, which forms the comparison axis for this piece.

Background

• Newswhoplus, “Seoul’s Foreign Domestic Care Worker Pilot Extended — Transition to Full Program Uncertain”, February 7, 2025. ··· Shows, from that point in time, why the plan to expand to 1,200 workers stalled.
• Labor Today, “‘Foreign Domestic Care Worker’ Pilot Extended by One Year”, February 14, 2025. ··· Lays out the extension decision alongside labor groups’ criticism of it.

Related issues

• The Country Where the Yogurt Lady Became a Social Safety Net — and the One Where She Became a Delivery Platform
• A Young Person Caring for a Child Isn’t “Not Economically Active”

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

📝 Glossary

각주

1. Gasa-sayongin (household employee): A person contracted directly by an individual household to perform domestic work. Under Article 11, Clause 1 of Korea’s Labor Standards Act, such workers are excluded from the law’s coverage. This is the exact fault line in the debate over whether minimum wage should apply to foreign domestic care workers. ↩

2. Employment Permit System (E-9): A system and visa status that lets workplaces unable to find domestic workers hire non-specialist foreign labor. The government sets annual quotas by industry. ↩

3. ILO Discrimination Convention: A convention adopted by the International Labour Organization to prohibit discrimination in employment and occupation. Ratifying countries may not set different wages based on nationality. Korea is a ratifying country. ↩