Of 4,279 Dismissed From Type A Support Providers, 936 Found Re-employment and 2,073 Moved to Type B — From Work With a Contract to Work Without One

Moving to Type B continuous employment support2,07348.4%

No employment contract; paid an allowance

Re-employment secured93621.9%

Work under an employment contract

Neither, as at end of August 20241,27029.7%

Between March and July 2024, Hello Work recorded 4,884 dismissals of people with disabilities, of whom 4,279 were users of Type A providers. Re-employment at 936 and moves to Type B at 2,073 together account for around 70%. The remainder is the balance; the material does not break it down.

4,279 people in five months

A document presented to the Social Security Council's disability committee in November 2024 sets out the figures.

Between March and July 2024, Hello Work recorded 4,884 dismissals of people with disabilities through dismissal notifications, of which 4,279 were users of Type A continuous employment support[1]. Five months.

Month by month there is a peak. The totals were 823 in March, 1,013 in April, 1,350 in May, 872 in June and 826 in July; the Type A share was 698, 887, 1,241, 766 and 687[1]. May was the highest, and in that month 91.9% of all dismissals of people with disabilities came from Type A sites.

The FY2024 fee revision took effect in April.

Twice as many moved to Type B as found new work

Where they went is set out. As of the end of August 2024, 936 had found new employment, and 2,073 had moved or were due to move to a Type B site[1], which the document says accounts for around 70% of the total.

The rest are 949 still job-seeking and 321 categorised as other[1]. Of the 949, some 932 are receiving job-search support at Hello Work[1]. The 321 include people whose intentions were undecided and people not seeking work.

Of the 936 who found work, 696 went to another Type A site

The document breaks down the 936. Of the 936 who found new employment, 696 took jobs at Type A sites[1].

That is 74.4% going to another Type A site. Those who left Type A altogether number 240, or 5.6% of the 4,279.

The 4,279 is not the total of Type A dismissals

A note gives the definition.

The number of dismissals at Type A continuous employment support sites is limited to dismissals at sites that filed notifications for 10 or more people within a single month. More than 90% of those sites run a deficit on production activity.

Sites that filed for nine or fewer in a month are not in this figure. The 4,279 is a floor for Type A dismissals, not the whole.

The same note describes the sites that were counted. More than 90% were running a deficit.

For scale: there were 4,472 Type A sites with 87,262 users[2].

The names are similar; the standing is not

Type A continuous employment support involves an employment contract. Labour law applies, including the minimum wage. Sites carry a condition, which the document states:

A designated Type A continuous employment support provider must ensure that the amount equivalent to the revenue from production activity less the expenses necessary for that activity is at least equal to the total wages paid to users.

In other words, what the business earns net of costs has to cover the wages. Benefit payments are not to fund them.

Type B involves no employment contract. What people receive is an allowance rather than a wage, and the minimum wage does not apply. In exchange, the conditions are looser and attendance can follow a person's health.

The gap is 4.9 times a month

The amounts are published. In FY2022 the average wage at Type A sites was 83,552 yen a month, against an average allowance of 17,031 yen at Type B sites[3]. A gap of 4.9 times.

The trend is published too. The Type A average rose from 68,691 yen in FY2012 to 86,752 yen in FY2023, up 18,061 yen over eleven years[4]. Type B went from 14,190 yen in FY2012 to 17,031 yen in FY2022, up 2,841 yen over ten years[5].

The FY2023 Type B figure is not a continuation of that line. It reads 22,649 yen, and the document leaves the year-on-year column blank, marked with a dash. It gives the reason.

For FY2023, the FY2024 fee revision changed the method of calculating the average monthly allowance, which has substantially increased the reported result.

The revision moved the score floor from 10 points to minus 65

Type A base fees are set by a score. The FY2024 fee revision reorganised the items. The document shows both versions.

Production activity now adds points where the balance exceeds total wages and subtracts where it falls below. A new deduction was added for sites not acting on a business improvement plan[1].

The maximum is 200 either way. The floor changed. It was 10 points before and reaches minus 65 after. Both items that push it down are about money.

Meanwhile varied ways of working and support capability each fell from a maximum of 35 to 15[1]. The two items measuring how users can work and how well staff can support them lost 40 points between them.

The look-back period lengthened too. The previous version assessed the past two years; the revised version adds a third[1].

The score moves the fee a long way. For a site with 20 or fewer places and a 7.5-to-1 staffing ratio, 170 points or more gives 791 units a day and under 60 points gives 325[1]. A spread of 2.4 times.

Half the sites were not meeting the standard to begin with

The document also shows the position before the revision. As of the end of March 2023, sites whose production revenue fell below the total wages paid numbered 1,882 out of the 3,715 whose finances were captured, or 50.7%[6]. The year before it was 1,984 of 3,512, or 56.5%[7].

That 50.7% is not a share of the 4,472 designated sites. Finances were captured for 3,715; the remaining 757, or 16.9%, were not.

Many had been in that position for a while. Of the 1,882, some 1,507 — 80.1% — had also failed the standard the previous year[1]. Business improvement plans had been filed by 1,690, a submission rate of 89.8%[1].

In four months Type A lost 2,844 users and Type B gained 16,053

User numbers are charted. Type A users fell from 90,106 in March 2024 to 87,262 in July 2024, down 2,844. Type B rose over the same months from 352,862 to 368,915, up 16,053[8].

Site counts moved the same way. Type A fell from 4,634 to 4,472, down 162, while Type B rose from 17,295 to 17,820, up 525[8].

It is the first fall in the Type A site count across a chart that starts in March 2015.

That said, the 2,073 who moved from Type A are 12.9% of Type B's increase of 16,053. Most of Type B's growth did not come from Type A.

The intent is legible

Reduce the sites that pay wages out of benefit funding and keep the ones that can earn. Where Type A is a place of employment, requiring it to work as a business follows. With 50.7% failing the standard and 80.1% of those failing it a second year running, the position was not one to leave alone either.

The question is where the people at the sites that did not work out end up.

The indicator being measured and the standing that moved are different things

The figures answer it. Against 936 finding new work, 2,073 moved to Type B. Of those who left Type A, about two in ten moved to work under an employment contract and close to five in ten moved to work without one.

And within that two in ten, 74.4% went to another Type A site. The 240 who moved to a workplace outside Type A are 5.6% of the 4,279.

Set against the policy's own direction, this runs backwards. Employment support for people with disabilities has been built around a path from welfare-based work toward open employment. Type A sits in the middle, holding an employment contract while receiving support, with progression to open employment as the intended next step.

What happened was that the middle rung was thinned by an assessment of business performance, and the people on it moved down.

The indicator that assesses a site and the indicator that describes a user's standing are not the same. The production balance measures the site's finances. The presence of an employment contract measures the user's position. The revision weighted the first heavily; how the second moves when the first deteriorates is not in the fee design at all.

The same revision also changed how Type B allowances are calculated

The FY2024 revision contained another change.

The average monthly allowance at Type B sites feeds into the fee schedule. Its denominator changed.

Up to FY2022 it was calculated using the previous year's number of allowance recipients as the denominator. In the FY2024 revision of disability welfare service fees, in consideration of sites accepting people who attend fewer days owing to the nature of their disability, a new method was introduced using the previous year's average number of daily users as the denominator.

The average number of daily users is smaller than the number of people receiving an allowance. A smaller denominator raises the per-person figure. It rises even when nobody's receipts change.

The ministry does not hide this. It cuts the line at FY2022, places FY2023 as a detached point and leaves the year-on-year column as a dash. Even so, a revision that measured the Type A side more harshly changed the arithmetic that makes the Type B side look better, in the same year and the same revision.

What the ministry did

Measures were taken. Prefectures were reminded of the points to observe when a Type A site closes, and in October 2024 were asked to strengthen coordination between designating and benefit-granting authorities and to share information with labour bureaux and Hello Work[1].

On the employment side there is information-sharing with local government to connect people facing dismissal to Hello Work, individual vocational counselling and placement, matching to vacancies for those judged able to move into open employment, group processing of unemployment insurance claims, and company tours and job fairs for people leaving Type A sites[1]. The 936 who found work are partly the result.

Even so, the 2,073 who moved to Type B are the larger number.

What is not known

What is known: how many were dismissed, and where they stood at the end of August. Three things are not.

The first is how many are missing from the 4,279. Sites that filed for nine or fewer in a month were not counted.

The second is what became of the 2,073. It is a snapshot at the end of August; whether they are still at Type B sites, or have since moved into open employment, is not shown.

The third is how their receipts changed. On the averages the gap is 4.9 times, but the individual amounts for these 2,073 are not published.

How the choice of indicator drives the result was covered in designing outcome indicators（このサイトの記事）; here the change happened on the side the indicator does not measure. The finances of the sites may well improve. The standing of the people who worked there moved without being measured.

The same revision cutting in opposite directions depending on the shape of the business appeared in 91 home-care bankruptcies against 45 in day services（このサイトの記事）. Where the fee unit sits decides which businesses survive.

On the employment side, an employment rate of 2.41%, a fourteenth straight record, with only 46.0% of firms meeting the quota（このサイトの記事） shows that half the firms on the receiving end fall short as well.

Further Reading

• 『介護格差』 (Care Inequality)（外部サイト、新しいタブで開きます） (Yasuhiro Yuki, Iwanami Shoten). An account, drawn from practice, of what separates those who can obtain welfare services from those who cannot. Useful for thinking about how fee design moves what happens on the ground.

References

On the State of Type A Continuous Employment Support (143rd Subcommittee on Persons with Disabilities, Document 3) — MHLW, Department of Health and Welfare for Persons with Disabilities and Disability Employment Measures Division (2024). Ministry of Health, Labour and Welfare

Social Security Council, Subcommittee on Persons with Disabilities — Ministry of Health, Labour and Welfare (2026). Ministry of Health, Labour and Welfare

Results on Allowances and Wages for Fiscal 2023 — MHLW, Department of Health and Welfare for Persons with Disabilities (2025). Ministry of Health, Labour and Welfare