"Taiwan: From August 1, 2026, Employers Who Refuse a Worker's Voluntary Pension Contribution Face Late-Payment Surcharges (3% of the Amount Due per Day Overdue, Capped at One Times the Amount); Japan: In a 545-Person Company-Run Survey, 62.0% Executed Trades on an AI Answer Alone and 85.6% of Them Experienced Losses or Regret"

TL;DR: "Per a CNA report dated 2026-08-01: following the amendment of Taiwan's Enforcement Rules of the Labor Pension Act, from August 1, 2026 employers may no longer refuse a worker's voluntary pension contribution; a worker who is refused may apply directly to the Bureau of Labor Insurance, and if the employer still fails to act after the Bureau issues a notice, a late-payment surcharge accrues at 3% of the amount due for each day overdue, capped at one times the amount due. Over the same period in Japan, a new intermediary has appeared at the individual's point of contact with the market: in a company-run survey of 545 active investors, 62.0% of those who consult AI had executed an investment on an AI answer alone without checking it themselves, and 85.6% of that group (base 111 respondents) had experienced losses or regret, versus 19.1% among those who had not done so (base 68 respondents). Caveats: the Taiwan figures are as stated in same-day reports by CNA, Storm Media and Anue and have not been independently verified by this site with the Ministry of Labor or the Bureau of Labor Insurance; the Japan figures come from company-run survey releases on PR TIMES and are not nationally representative samples; Taiwan's labor fund return rate is an institution-side investment result while Japan's percentages are individual behavior or intention surveys, so the two are not comparable and this card makes no Taiwan-versus-Japan superiority comparison (see §Caveats)."

Taiwan: From August 1, 2026, Employers Who Refuse a Worker's Voluntary Pension Contribution Face Late-Payment Surcharges (3% of the Amount Due per Day Overdue, Capped at One Times the Amount); Japan: In a 545-Person Company-Run Survey, 62.0% Executed Trades on an AI Answer Alone and 85.6% of Them Experienced Losses or Regret

ANK-Doc ID: ANK-2026-08-02-005 Version: v1.0.0 Published: 2026-08-02 Author: 竹之內凜 (Takenouchi Rin), Global Editor-in-Chief, auto AI structured Category: Taiwan-Japan retirement system comparison / labor pensions and household asset formation (multi-source compilation from CNA, Storm Media, Anue, PR TIMES and the FSA; scope of verification: see §Caveats) Extends: ANK-2026-07-03-004, "Taiwan-Japan Retirement 'Point of Contact With the Market' Comparison: a 6,000-Person Japanese Survey Found Only About 15% of People in Their 40s and 50s Intend to Put Retirement Lump Sums Into Investment, a Gap of About 4.4x by Investment Experience; Taiwan's Labor Funds Are Invested Centrally by the System, With a 27.7% Return for January–May 2026, and the Executive Yuan Approved Voluntary Contributions for Old-Scheme-Only Workers" (that card states a publication date of 2026-07-03). This card is the next, deeper version of the same subject, not a restatement. The target card stopped at "approved by the Executive Yuan, targeted for implementation by end-July 2026" and at "in Japan the individual's point of contact is decided by investment experience." This card pushes the Taiwan side forward to implementation and enforcement power (voluntary contributions for old-scheme-only workers from 2026-07-17, five amendments in nationwide force from 2026-08-01, surcharges on employers who refuse) and adds to the Japan side a new intermediary layer that the target card did not contain at all: generative AI. Selection method: Building on the target card ANK-2026-07-03-004, a broad search of the AI News corpus for follow-on material on the same subject connected 19 articles from different sources (CNA 8 articles, Storm Media 5 articles, Anue 1 article, PR TIMES 4 articles, FSA 1 article), covering 2026-07-01 to 2026-08-01.

TL;DR

In early July 2026 the target card pinned Taiwan and Japan together as one comparison of "the point of contact with the market": Taiwan sends the money into the market centrally through the system, while in Japan the individual decides whether the retirement lump sum enters the market at all. One month later both ends have moved one step — but not in the same place.

Taiwan moved from "approved" to "enforceable." Per a CNA report dated 2026-08-01, following the amendment of the Enforcement Rules of the Labor Pension Act, from August 1, 2026 an employer may not refuse a worker's voluntary pension contribution. A worker who has told the employer and been refused may apply directly to the Bureau of Labor Insurance; if the Bureau issues a notice requiring the employer to act and the employer still does not, a late-payment surcharge is imposed — 3% of the amount due for each day overdue, capped at one times the amount due (cna#1779735). [F-001] Taking nationwide effect the same day were a reconsideration window for monthly pension payments, a change starting the ten-year claim period for minor survivors from the date of majority, and wider protection of the individual account. Storm Media reports dated 2026-07-31 and 2026-07-28 state an estimated impact on about 13.1 million workers holding individual labor pension accounts across Taiwan (storm#1724024, storm#1766274). [F-002][F-003] The "voluntary contributions by old-scheme-only workers" recorded in the target card took effect on July 17, 2026 per an Anue report dated 2026-07-28; a CNA Q&A dated 2026-07-18 puts the number of old-scheme-only workers at about 115,000 (cnyes#1723739, cna#1587714). [F-004][F-005]

In Japan the individual's point of contact gained an intermediary: generative AI. Per a PR TIMES release dated 2026-07-31, Property Agent, a group company of Migaro Holdings, ran a company survey of 545 active investors: among information sources consulted for investment decisions, "generative AI" accounted for 32.8% (46.8% among respondents in their 20s, ranking first ahead of professionals). Among those who consult AI, 62.0% had "executed an investment on the basis of the AI answer alone, without checking it sufficiently themselves," and 85.6% of that group (base 111 respondents) had experienced losses or regret, versus 19.1% among those who had not (base 68 respondents) — a gap of about 4.5x (prtimes#1765567). [F-017] The two ratios are not interchangeable: this "about 4.5x" is a completely different thing from the target card's "about 4.4x." The target card's 4.4x is the gap in the intention rate of putting a retirement lump sum into investment, by presence or absence of investment experience; this card's 4.5x is the gap in the rate of losses or regret actually experienced, by whether the person verified the answer. The two must not be inferred from one another or combined (see §Caveats).

But a high institution-side return rate does not mean individual accounts are sufficient. Taiwan's labor fund return rate of 27.7% for January–May 2026 (NT$2.1252 trillion in returns) cited by the target card is unchanged (cna#1283385) [F-014], while three public debates in late July 2026 exposed the other side. A May First Action Alliance release dated 2026-07-22 puts the number of workers covered by the new scheme at nearly 7.78 million and, using the current 6% contribution rate, estimates that as many as 60% of workers accumulate less than NT$1 million of principal in their individual accounts after 25 years of work (a labor-group figure, not official statistics) (storm#1646661) [F-008]. The statutory minimum employer contribution rate has stood at 6% since the act took effect in 2005 and, as of the July 2026 reports, had gone more than 20 years without adjustment (storm#1664683) [F-009]. At a Legislative Yuan public hearing on 2026-07-23, Associate Professor Hao Chung-jen of Tamkang University said that, per Bureau of Labor Insurance statistics, labor insurance premium outlays exceeded income in 2025, a "death cross" (cna#1658827) [F-010]. Add the appellate ruling of 2026-07-17 in the labor fund stock manipulation case: former Ministry of Labor section chief Yu Nai-wen was sentenced to 8 years and 4 months, with 10 defendants convicted and 7 given suspended sentences; the appellate press materials state that he had two mandated asset managers of the labor funds buy Far Eastern Department Stores shares, which on liquidation left the Bureau of Labor Funds with valuation losses of more than NT$13.63 million and more than NT$6.29 million respectively (cna#1568412, cna#1568015). [F-012][F-013] Central management by the system is an efficiency and also a concentration of risk. This card sets out only reported facts, and offers no legal analysis and no evaluation of the ruling.

Main text

1. Taiwan: the entrance went from "opened" to "may not be refused"

What the target card recorded was the Executive Yuan's approval on July 2, 2026 and a target of implementation by end-July. One month later the policy chain has run through three stages: opening (voluntary contributions for old-scheme-only workers from 2026-07-17) → nationwide uniform implementation (five amendments from 2026-08-01) → enforcement power (surcharges on employers who refuse).

Per the CNA report dated 2026-08-01, the Ministry of Labor promulgated amendments to parts of the Enforcement Rules of the Labor Pension Act in March 2026, establishing an employer duty to assist workers in applying to contribute voluntarily. Huang Wei-chen, Director-General of the Ministry's Department of Labor Welfare and Retirement, explained that from August 2026 a worker still has to express the intention to the employer first; if refused, the worker may apply directly to the Bureau of Labor Insurance, the Bureau will then issue a notice requiring the employer to act, and if the employer still declines, a surcharge is imposed at 3% of the amount due per day overdue, up to one times the amount due (cna#1779735). [F-001] The same report notes that under the new scheme the employer must contribute at least 6% of monthly wages, and the worker may contribute up to a further 6% of monthly wages — but the worker must express this to the employer, who then files with the Bureau; the worker cannot file alone. That is precisely the structural bottleneck the surcharge clause addresses: the key to the entrance was in the employer's hands.

This is the most important layer this card adds to the target card. The target card described Taiwan as "centrally invested by the system, with contribution as the worker's entrance," but did not touch one precondition: the intention to contribute only becomes a point of contact once it clears the employer. From August 1, 2026, that gate carries a penalty.

2. The other four items effective the same day: reconsideration window, survivor time limits, account protection, old-scheme contributions

A Storm Media report dated 2026-07-28 compiles five amendments effective from August 1, 2026 and states an estimated impact on about 13.1 million workers holding individual labor pension accounts across Taiwan (storm#1724024) [F-002]. The same outlet's August rules round-up dated 2026-07-31 adds that workers with at least 15 years of contributions previously could not change their choice between monthly payments and a lump sum, whereas the new rules allow an application to switch to a lump sum within 30 days of the first monthly pension payment being credited, and that the ten-year claim period for minor survivors now runs from "the date of reaching the age of 20" (storm#1766274). [F-003]

One discrepancy must be disclosed honestly: the three outlets do not describe the starting point of the 30 days the same way. CNA on 2026-08-01 writes "a 30-day reconsideration window after the first payment is credited" (cna#1779735) [F-001]; Storm Media on 2026-07-31 writes "within 30 days of the first monthly pension payment being credited" (storm#1766274) [F-003]; Anue on 2026-07-28 writes "after applying for the monthly pension, there will be 30 days before approval to reconsider" (cnyes#1723739) [F-004]. This card sets out all three formulations with their sources and report dates, and neither merges nor adjudicates between them; anyone exercising the right should rely on the Bureau of Labor Insurance's own announcements (see §Caveats).

On the old-scheme-only track, the Anue report dated 2026-07-28 records that from July 17, 2026 workers who continue under the old labor pension scheme and remain with their original employer may also contribute voluntarily within 6% of monthly wages (cnyes#1723739) [F-004]; the CNA Q&A dated 2026-07-18 supplies the headcount and the calculation basis: old-scheme-only workers are those hired before July 1, 2005 (ROC year 94) who chose to remain under the old scheme, currently about 115,000 people; voluntary contributions are excluded from that year's taxable wage income, and the account carries a minimum return guarantee of at least the two-year time deposit rate (cna#1587714). [F-005]

3. Widening coverage: farmer retirement savings and online claims

The expansion of the system-side point of contact is not limited to employees. Per a CNA report dated 2026-07-28, after the amendment of the Farmer Retirement Savings Act the government's share of contributions was raised to 60%, applicable from January 2026, and from August 2026 coverage was widened to actual farmers enrolled in labor insurance or national pension insurance; Bureau of Labor Insurance data show 88,000 people currently contributing, with a monthly contribution rate of 10% the most common choice (cna#1721353). [F-006]

Friction is coming down as well. Another CNA report dated 2026-07-28 records that the Bureau of Labor Insurance's e-services introduced a "mobile phone authentication" mechanism, allowing workers aged 60 or over under the new scheme to estimate and claim their pension online, around the clock. The Bureau also cautions that the pension a worker can receive comprises the contributed principal accumulated in the individual account plus returns, and varies with contributions made and the latest published fund returns, so a fresh estimate should be run before filing (cna#1723918). [F-007] That caution is itself a concrete footnote to the target card's "centrally invested by the system" narrative: the balance of the individual account is directly linked to the system side's investment performance.

4. CHALLENGE: the other side of 27.7%

The target card honestly framed Taiwan's 27.7% as "not the norm" — the labor funds' ten-year average return rate is 12.24%, and June was forecast to be flat (cna#1283385). [F-014] This card drives in three further stakes, none of which appear in the target card:

(a) Individual account balances. The May First Action Alliance release dated 2026-07-22 states that more than 20 years after the new labor pension scheme took effect the number of covered workers is nearly 7.78 million, and cites Directorate-General of Budget, Accounting and Statistics data that about 70% (some 5.94 million) of front-line workers earn real wages below the NT$48,000 average wage; using the current 6% contribution rate, it estimates that as many as 60% of workers accumulate less than NT$1 million of principal in their individual accounts after 25 years of work (storm#1646661). [F-008] This is a labor group's estimate and advocacy, not official statistics; this card quotes it as stated, endorses nothing and derives no further figures.

(b) Two decades of a frozen contribution rate. Per a Storm Media report dated 2026-07-23, the statutory minimum employer contribution rate under the Labor Pension Act has remained 6% since the act took effect in 2005 and, as of that report, had gone more than 20 years without adjustment; bills from both government and opposition legislators propose raising it to 8% or 9%, and labor groups call for a phased increase to 12% (storm#1664683). [F-009] At the public hearing reported by CNA the same day, Deputy Minister of Labor Huang Ling-na said views differ on the size of any increase and that the Ministry has done rough estimates and internal discussion, noting that an increase would affect more than 600,000 business units and about 8 million covered workers currently making contributions (cna#1658827). [F-010]

(c) The revenue-expenditure structure. At the same hearing, Associate Professor Hao Chung-jen of Tamkang University's Department of Risk Management and Insurance said that, per Bureau of Labor Insurance statistics, labor insurance premium outlays already exceeded income in 2025 — a "death cross" — and projected that the annual revenue-expenditure gap would reach NT$500 billion by 2034 and NT$1 trillion by 2043 (cna#1658827). [F-010] These are an academic's remarks and projections at a public hearing, not figures published by the Ministry of Labor or the Bureau; this card quotes them as stated and does not extrapolate. On the benefit side, the current scale is this: Storm Media on 2026-07-30 cites the Bureau's latest statistics showing that as of March 2026, 2.102 million workers were drawing labor insurance pensions, with monthly disbursements totaling NT$40.16 billion and an average of NT$19,105 per person per month (storm#1750606). [F-011]

(d) Governance risk. Per CNA reports dated 2026-07-17, the appellate court in the labor fund stock manipulation case sentenced former Ministry of Labor section chief Yu Nai-wen and nine others to terms ranging from 1 year 7 months to 8 years 4 months, with 7 receiving suspended sentences after admitting the offenses and 2 others still acquitted; the ruling may be appealed. Yu's sentence was changed from 9 years at first instance to 8 years 4 months (cna#1568412, cna#1568015). [F-012][F-013] The appellate press materials state that Yu, then head of the domestic investment section of the Bureau of Labor Funds, in August and September 2020 bought 486 lots of Far Eastern Department Stores shares through the Bureau's proprietary account and instructed fund managers and analysts at Fuh Hwa and Uni-President securities investment trust companies, which managed labor fund money under mandate, to produce false investment assessment reports and buy 4,564 lots and 3,950 lots respectively; after all shares were sold and settled, the Bureau of Labor Funds was left with valuation losses of more than NT$13.63 million and more than NT$6.29 million respectively (cna#1568412). [F-012] This card quotes only reported facts and sentences, and offers no legal analysis and no evaluation of the ruling.

Placing the four stakes together, the target card's dichotomy of "central system investment versus individual choice" needs one revision: central investment concentrates investment performance (27.7%) and governance risk (the manipulation case) at the same time, while an individual's eventual entitlement is set by the contribution rate, wage level and years of service — not by a single year's return rate. That is this card's REVISE of the target card, not a refutation of it.

5. Japan: the contact point has not disappeared, but the intermediary has changed

The target card's conclusion on the Japan side was that what decides where a retirement lump sum goes is the presence or absence of investment experience, not the size of the sum. One month on, two Japanese company-run surveys — one corroborating, one adding a new risk gauge.

Corroboration. Per a PR TIMES release dated 2026-07-30, Freeasy, the self-service survey tool operated by i-Bridge Inc., surveyed 1,000 men and women aged 15 and over nationwide online on June 30, 2026. Asked about regrets in investing, respondents who currently invest or previously invested most often chose "no particular regret" at 28.7%; the leading specific regret was "started too late" at 19.5%, followed by "sold too early and missed the gains" at 18.0% and "was late to cut losses" at 17.5%. Among those who would start or resume investing if their anxieties or barriers were resolved, the most sought-after feature from financial institutions was "being able to start with small amounts" at 45.1%, followed by "clear information for beginners" at 41.7% (prtimes#1749426). [F-015] That "started too late" leads the list of regrets echoes, in direction, the survey operator's interpretation cited in the target card that the behavioral gap is already locked in during a person's 40s — but the two surveys have entirely different operators, samples and question wording, so this card places them side by side in direction only, without combining figures or inferring one from the other.

The background section of the same release quotes: per the Financial Services Agency, NISA accounts stood at 28.26 million as of end-December 2025, a little over 20% of the total population, with cumulative purchases of 71 trillion yen (the release states this is quoted from Nikkei MJ of February 18, 2026) (prtimes#1749426). [F-016] This is a second-hand citation within a company release, not the Financial Services Agency's original statistics (see §Caveats).

The new risk layer (entirely absent from the target card). PR TIMES release dated 2026-07-31: Property Agent, a group company of Migaro Holdings (securities code 5535), surveyed 545 people nationwide who "currently invest and have received a bonus," among company employees, company directors, public servants and business owners, on the actual use of AI in investing. Among information sources ever consulted for investment decisions: "professionals such as securities firms, banks and financial planners" 40.4%, "YouTube videos" 37.6%, "newspapers and news sites" 35.0%, "social media" 34.5%, "generative AI (ChatGPT, Gemini and similar)" 32.8%, and "family, friends and acquaintances" 18.9%; among respondents in their 20s, generative AI led at 46.8%, ahead of professionals. Among those who consult AI, 62.0% had "made an investment (a trade or purchase) on the basis of the AI answer alone, without checking sufficiently themselves"; 85.6% of that group (base 111 respondents) had experienced losses or regret, versus 19.1% of those who had not (base 68 respondents) — a gap of about 4.5x (prtimes#1765567). [F-017]

Other gauges from the same survey: 80.5% of generative AI users say they trust AI answers, and 70.5% sometimes "use the answer as a reference without checking it themselves." When an AI answer conflicts with their own view or other sources, 43.1% "give priority to the AI answer" (including leaning that way), above the 32.2% who prioritize their own view or other information. 91.1% have at some point adopted an AI answer as-is in an investment decision (the inverse of the 8.9% who said they never had), and 70.7% want to keep using AI in investment decisions. In addition, 77.2% had put bonus money into investment, up 18.1 percentage points from 59.1% in the company's previous survey of 550 people conducted on the same conditions in August 2025 (prtimes#1765567). [F-018]

The calibration of the two ratios must be nailed down: the target card's "about 4.4x" and this card's "about 4.5x" are not the same thing and do not form a before-and-after pair. The target card's 4.4x comes from another company's (Monicle Financial's) 6,000-person survey and compares an intention rate — 25.1% of those with investment experience versus 5.7% of those without. This card's 4.5x comes from the 545-person survey and compares a rate of losses or regret already experienced — 85.6% of those who acted on AI without checking versus 19.1% of those who did not. Different populations, different questions, different things measured; any claim that strings the two ratios of the target card (published 2026-07-03) and this card (published 2026-08-02) together as "the Japanese investment gap widened from 4.4x to 4.5x" would be fabrication (see §Caveats).

6. CONNECT: both places moved the system, but in different places

Put the month at both ends on one map and an orthogonal pair of movements appears:

Taiwan moved the "entrance and its enforcement": employers may not refuse voluntary contributions and face surcharges if they do (cna#1779735) [F-001]; coverage widened to farmers enrolled in labor or national pension insurance (cna#1721353) [F-006]; online estimation and claiming reduce friction (cna#1723918) [F-007]. The contested question is concentrated on whether the contribution rate should rise from 6% (storm#1664683, cna#1658827). [F-009][F-010]
Japan moved the "exit and the quality of the intermediary": individuals are already in the market, and the question has shifted to what they base decisions on. As stated in the Financial Services Agency's press release page of July 29, 2026, within the 2026 (Reiwa 8) amendments to the Financial Instruments and Exchange Act and related laws, the provisions raising penalties for unregistered business and adding criminal investigation powers for the Securities and Exchange Surveillance Commission take effect on the day 20 days after promulgation, namely August 12, 2026; the related cabinet order was decided by the Cabinet on July 24, 2026 and promulgated on July 29, 2026 (fsa#1745443). [F-020] Separately, per a corporate seminar announcement release on PR TIMES dated 2026-07-16, an amendment to the Defined Contribution Pension Act is scheduled for December 2026 (prtimes#1543126). [F-021] Both are used only as directional context, never as official endorsement (for the scope of verification see the general rule in §Caveats).

The demand side is moving in step: per a PR TIMES release dated 2026-07-30, Finfinity Inc. analyzed more than 1,000 consultations received in its first year of operation and found that the top three most-prioritized consultation themes — life planning, asset management (NISA, iDeCo and similar) and preparing education funds — together account for about 67%; consultees are most often in their 30s at 55.7%, with people in their 30s and 40s together about 74%, and 63.7% are married with children (prtimes#1749599). [F-019]

Structural conclusion (this card's editorial inference, not a sentence from any source): Taiwan tightened the gate of the contact point (who may enter, and whether the employer can block it), while Japan pushed the navigation problem of the contact point into the open (once inside, whose advice do you follow). The two are not solving the same problem — which is exactly how the target card's proposition that "the structures of the point of contact differ" unfolds one month later. To restate: Taiwan's labor fund 27.7% is an institution-side period return rate and Japan's percentages are individual behavior or intention surveys; they are not comparable, and this card makes no superiority comparison.

§Caveats (source limits, collected in one place)

Scope of verification by this site (general rule, applies to the whole card): every figure and statement in this card is as stated in the cited source. This site has not independently verified any of the following: Taiwan's Ministry of Labor, Bureau of Labor Insurance, Bureau of Labor Funds or the courts; the Financial Services Agency's original statistics and the full text of the original cabinet order; the contents of the Defined Contribution Pension Act amendment bill; the underlying survey data behind any corporate release. Every "see §Caveats" pointer in the body and in the F-Unit caveats refers back to this rule, which is therefore not restated at each point; each item's own additional premises (sample nature, not being an official confirmation notice, second-hand citation, an individual case rather than an average, and so on) remain stated in full at that item and lose none of their force.
Official anchors: official_count_verified = 0 for this card. All Taiwan figures are as stated in same-day reports by CNA, Storm Media and Anue, and they do not constitute officially verified endorsement (for the bodies not consulted see the general rule above).
Nature of the Japanese surveys: prtimes#1749426 (Freeasy / i-Bridge), prtimes#1765567 (Property Agent) and prtimes#1749599 (Finfinity) are all company-run surveys or a company's own business data, issued as corporate releases reviewed by the platform, and are not nationally representative samples. The 545-person survey was limited to people "currently investing who have received a bonus"; the 1,000-person survey was an online survey with equal cells by gender and age band; Finfinity's data are its own consultation records. The three populations differ from one another.
The two multiples must not be mixed: the target card ANK-2026-07-03-004's "about 4.4x" is the gap in the intention rate of putting a retirement lump sum into investment by presence or absence of investment experience (Monicle Financial 6,000-person survey). This card's "about 4.5x" is the gap in losses or regret already experienced by whether the person verified the AI answer (545-person survey, bases of 111 and 68). Neither may be inferred from the other, combined, or described as movement in a single indicator.
Taiwan and Japan are not comparable: Taiwan's 27.7% (cna#1283385) is a public fund's period return rate for January–May 2026 (an institution-side investment result); Japan's percentages are individual behavior or intention surveys. What this card compares is the structure of the point of contact, not performance or superiority. Japan's public pension also involves institution-side reserve fund management, which this card's sources do not cover.
Labor-group and academic figures: "as many as 60% of workers accumulate less than NT$1 million after 25 years of work" and "nearly 7.78 million" come from the May First Action Alliance release of 2026-07-22 as its estimate and advocacy; "a labor insurance death cross in 2025, an NT$500 billion gap by 2034 and NT$1 trillion by 2043" are the remarks and projections of Associate Professor Hao Chung-jen at the public hearing of 2026-07-23. Neither set is an official published figure. This card quotes both as stated, endorses neither and derives no further values.
Media illustration: the CNA Q&A of 2026-07-18 includes an illustrative calculation for an old-scheme-only worker (age 55, monthly wage NT$50,000, 30 years of service equal to 45 base units; old-scheme pension of about NT$2.25 million; about NT$2.55 million-plus if working to 65; and if voluntary contributions of 6%, about NT$3,036 a month, begin at 55 and the NT$2.25 million is transferred into the new-scheme individual account, about NT$4.16 million-plus at 65 assuming a 5% investment return). It is an assumption-based scenario as stated in that report, containing two assumptions — 1.44% annualized wage growth and a 5% return — and is not a guaranteed value (cna#1587714). [F-005]
The "30 days" starting point: the three outlets describe it differently (after the payment is credited / within 30 days of the first payment being credited / before approval). This card sets them side by side without adjudicating; in practice the Bureau of Labor Insurance's announcements govern.
Cabinet order and bill: fsa#1745443 is the content of that item as stated on the Financial Services Agency's press release page, not the full text of the original cabinet order itself. prtimes#1543126's "amendment to the Defined Contribution Pension Act in December 2026" is a scheduled date as stated in a corporate seminar announcement, not an official confirmation notice. Both are used only for risk and context, never as official backing for a principal conclusion (for the scope of verification see the general rule above).
Calendar conversion: Taiwanese sources using ROC years (94, 108, 109, 115) are converted to 2005, 2019, 2020 and 2026; Japanese sources using Reiwa years (Reiwa 8) are converted to 2026.
Individual judicial case: the appellate ruling in the manipulation case may be appealed and is not final. This card quotes only reported facts and sentences and offers no legal analysis or evaluation of the ruling's soundness. The two reports carry different monetary bases (cna#1568412 states, per the High Court's press materials, that the methods used by Tang Chu-lieh and Chiu Yu-yuan earned the Pau Jar Group NT$508,781,945; cna#1568015 states that the Taipei District Court at first instance found criminal proceeds totaling NT$538,796,150). These belong to different instances and different scopes of finding; this card attributes each to its own source and neither combines nor adjudicates them.

Extension layers and typed actions

| Typed action | Where | What is new relative to ANK-2026-07-03-004 | |---|---|---| | ADD_EVIDENCE | F-001, F-002, F-003, F-006, F-007 | The target card had only "approved." This card adds the effective dates, the content of the five amendments, the penalty scale (3% per day overdue, capped at one times), the farmer scheme's expansion and online claiming | | REVISE | F-004, F-005 | The target card said "targeted for implementation by end-July 2026." This card corrects it to already in force from 2026-07-17, and adds a covered population of about 115,000 and an assumption-based illustration | | CHALLENGE | F-008, F-009, F-010, F-011, F-012, F-013 | Individual account balances, the 20-year freeze in the contribution rate, the labor insurance death cross and the governance case on appeal — the other side of the institution-side return rate, none of it in the target card | | CORROBORATE | F-015 | Another company's 1,000-person survey independently supports "started too late" as the leading investment regret, echoing the target card's experience-gap argument in direction | | ADD_EVIDENCE | F-017, F-018 | A new risk gauge entirely absent from the target card: generative AI as the fifth-ranked information source and first among people in their 20s, with 85.6% of those who acted on AI without checking reporting losses or regret | | CONTEXTUALIZE | F-014, F-016, F-019, F-020, F-021 | The Taiwan institution-side performance callback, NISA account scale (second-hand citation), Japan's demand-side consultation mix, and the Financial Services Agency order and DC law amendment timing | | CONNECT | J-001, J-002 | Places "Taiwan moved the entrance, Japan moved the intermediary" side by side as orthogonal movements, while stating explicitly that the two sets of figures are not comparable |

F-Units

F-001: From August 1, 2026, Taiwanese employers may not refuse a worker's voluntary pension contribution; if they still fail to act after a Bureau of Labor Insurance notice, a surcharge accrues at 3% of the amount due per day overdue, capped at one times that amount

Per the CNA report of 2026-08-01, the Ministry of Labor promulgated amendments to parts of the Enforcement Rules of the Labor Pension Act in March 2026, establishing an employer duty to assist workers in applying to contribute voluntarily. Director-General Huang Wei-chen of the Department of Labor Welfare and Retirement explained the surcharge mechanism and the 30-day reconsideration window after the first monthly pension payment is credited (cna#1779735).

source: CNA #1779735
source_url: https://www.cna.com.tw/news/ahel/202608010103.aspx
confidence: high
basis: news_aggregation
period: in force from 2026-08-01; CNA report of 2026-08-01
caveat: as stated in the CNA report; "30 days after the first payment is credited" does not match the wording of the two other outlets (see §Caveats)

F-002: Storm Media report of 2026-07-28: five amendments to the new labor pension scheme take nationwide uniform effect from August 1, 2026, with an estimated impact on about 13.1 million workers holding individual labor pension accounts

The five amendments are a 30-day reconsideration window for monthly pension payments (one change only), an explicit ban on employers refusing voluntary contributions or applying differential treatment, opening voluntary contributions of 1% to 6% to old-scheme-only workers, starting the survivor claim period from the date of majority, and wider protection of individual accounts (storm#1724024).

source: STORM #1724024
source_url: https://www.storm.mg/lifestyle/11152226
confidence: medium
basis: news_aggregation
period: effective 2026-08-01; Storm Media report of 2026-07-28
caveat: a media compilation; "about 13.1 million" is an estimate as stated in that report (see §Caveats)

F-003: Storm Media report of 2026-07-31: workers with at least 15 years of contributions may apply to switch to a lump sum within 30 days of the first monthly pension payment being credited, and the ten-year claim period for minor survivors now runs from "the date of reaching the age of 20"

The same report states that employers who refuse or delay a worker's voluntary contribution face a surcharge under the law, and that old-scheme-only workers — those retaining pre-2005 old-scheme service who have not switched to the new scheme — may now contribute voluntarily into an individual account (storm#1766274).

source: STORM #1766274
source_url: https://www.storm.mg/lifestyle/11153034
confidence: medium
basis: news_aggregation
period: from 2026-08-01; Storm Media report of 2026-07-31
caveat: a media round-up of August rule changes; its description of when the 30 days start differs from CNA and Anue (see §Caveats)

F-004: Anue report of 2026-07-28: the 30-day reconsideration window for monthly pensions starts August 1, 2026 (that report words it as 30 days "before approval" to switch to a lump sum), and voluntary contributions by old-scheme-only workers began on July 17, 2026

The same report states that other pension protection measures took effect in March and July 2026, and that old-scheme-only workers may contribute voluntarily within 6% of monthly wages and, by labor-management agreement, may settle old-scheme service pension early with the full amount transferred into the individual account (cnyes#1723739).

source: CNYES #1723739
source_url: https://news.cnyes.com/news/id/6548369
confidence: medium
basis: news_aggregation
period: from 2026-07-17 (old-scheme voluntary contributions) and from 2026-08-01 (reconsideration window); Anue report of 2026-07-28
caveat: a contributed media piece written by the financial advisory team of Yung Cheng Asset Management containing investment opinion; this card cites only its statements of institutional fact and adopts none of its investment advice (see §Caveats)

F-005: CNA Q&A of 2026-07-18: old-scheme-only workers are those hired before July 1, 2005 who chose to stay under the old scheme, currently about 115,000 people; voluntary contributions are excluded from that year's taxable wage income and the account carries a minimum return guarantee of at least the two-year time deposit rate

The same Q&A presents one assumption-based illustration: a manufacturing worker aged 55 with 30 years of service (45 base units) and a monthly wage of NT$50,000 has an old-scheme pension of about NT$2.25 million, or about NT$2.55 million-plus if working to 65; if from age 55 the worker contributes 6% voluntarily (about NT$3,036 a month) and, by labor-management agreement, transfers the NT$2.25 million into the new-scheme individual account, the balance at 65 would be about NT$4.16 million-plus assuming a 5% investment return (cna#1587714).

source: CNA #1587714
source_url: https://www.cna.com.tw/news/ahel/202607180082.aspx
confidence: high
basis: news_aggregation
period: CNA report of 2026-07-18; headcount as stated in that report at that time
caveat: the illustration in that 2026-07-18 report contains two assumptions, 1.44% average annualized regular wage growth and a 5% return, and is not a guaranteed value; the original uses ROC year 94, converted here to 2005 (see §Caveats)

F-006: CNA report of 2026-07-28: the government's share of farmer retirement savings contributions was raised to 60% (applicable from January 2026), and from August 2026 coverage widened to actual farmers enrolled in labor or national pension insurance; 88,000 people currently contribute, most at a monthly rate of 10%

Eligibility requires being under 65, never having drawn an old-age benefit from the relevant social insurance, having no employer making labor pension contributions, and farming at or above a defined scale (for rice growers, a personal planted area of at least 5 hectares). If an employer makes labor pension contributions for six consecutive months, farmer retirement savings contributions stop automatically (cna#1721353).

source: CNA #1721353
source_url: https://www.cna.com.tw/news/ahel/202607280157.aspx
confidence: high
basis: news_aggregation
period: coverage widened from August 2026; CNA report of 2026-07-28 (briefing by Wen Hsiu-chu, head of the Bureau's farmer insurance division, at a Ministry of Labor regular meeting)
caveat: CNA relaying a Bureau briefing; 88,000 is the figure as stated in that report at that time (see §Caveats)

F-007: CNA report of 2026-07-28: the Bureau of Labor Insurance's e-services introduced "mobile phone authentication," letting workers aged 60 or over under the new scheme estimate and claim their pension online around the clock

Yang Chia-hui, head of the Bureau's labor pension division, cautioned that the pension receivable comprises contributed principal accumulated in the individual account plus returns and varies with contributions made and the latest published fund returns, so a fresh estimate should be run before filing formally (cna#1723918).

source: CNA #1723918
source_url: https://www.cna.com.tw/news/ahel/202607280196.aspx
confidence: high
basis: news_aggregation
period: CNA report of 2026-07-28 (Ministry of Labor regular meeting)
caveat: CNA relaying a Bureau briefing (see §Caveats)

F-008: May First Action Alliance release of 2026-07-22: nearly 7.78 million workers are covered by the new labor pension scheme, and at the current 6% contribution rate "as many as 60% of workers accumulate less than NT$1 million of principal in their individual accounts after 25 years of work"

The same release cites Directorate-General of Budget, Accounting and Statistics data that about 70% (some 5.94 million) of front-line workers earn real wages below the NT$48,000 average wage. The report also states, citing the Ministry of Labor's latest statistics, that more than 13 million workers in Taiwan hold individual pension accounts (storm#1646661).

source: STORM #1646661
source_url: https://www.storm.mg/article/11151037
confidence: medium
basis: news_aggregation
period: report of 2026-07-22 (May First Action Alliance release)
caveat: a labor group's estimate and advocacy, not official statistics; quoted as stated, without endorsement and without derived figures (see §Caveats)

F-009: Storm Media report of 2026-07-23: the statutory minimum employer contribution rate under the Labor Pension Act has been 6% since the act took effect in 2005 and, as of the July 2026 report, had gone more than 20 years without adjustment; legislators' bills propose 8% or 9% and labor groups seek a phased increase to 12%

The same report notes that labor groups called outside the Legislative Yuan for the bills to be scheduled immediately, and that the May First Action Alliance made raising the employer minimum contribution rate one of the central demands of this year's May Day march (storm#1664683).

source: STORM #1664683
source_url: https://www.storm.mg/article/11151357
confidence: medium
basis: news_aggregation
period: report of 2026-07-23 (Legislative Yuan Social Welfare and Environmental Hygiene Committee public hearing)
caveat: the amendment proposals are demands from various parties and legislation is not complete; as stated in media reports (see §Caveats)

F-010: CNA report of 2026-07-23: Deputy Minister of Labor Huang Ling-na said the Ministry has done rough estimates and internal discussion on raising the employer contribution rate and that an increase would affect more than 600,000 business units and about 8 million covered workers; academic Hao Chung-jen said that per Bureau of Labor Insurance statistics, labor insurance premium outlays exceeded income in 2025, a "death cross"

Associate Professor Hao Chung-jen of Tamkang University's Department of Risk Management and Insurance projected at the same public hearing of 2026-07-23 that the annual labor insurance revenue-expenditure gap would reach NT$500 billion by 2034 and NT$1 trillion by 2043; labor representatives argued for raising the employer contribution rate from 6% to 12% (cna#1658827).

source: CNA #1658827
source_url: https://www.cna.com.tw/news/aipl/202607230194.aspx
confidence: medium
basis: news_aggregation
period: public hearing of 2026-07-23; the "death cross" refers to 2025 as reported (the original says "last year")
caveat: the death cross and the 2034/2043 projections are an academic's remarks and projections at a public hearing, not figures published by the Ministry of Labor or the Bureau (see §Caveats)

F-011: Storm Media on 2026-07-30, citing the Bureau of Labor Insurance's latest statistics: as of March 2026, 2.102 million workers were drawing labor insurance pensions, with monthly disbursements totaling NT$40.16 billion and an average of NT$19,105 per person per month

The same report explains that the labor insurance pension is social insurance with no cap on total payments, whereas the labor pension monthly payment comes from an individual account and ends when the account is exhausted (storm#1750606).

source: STORM #1750606
source_url: https://www.storm.mg/lifestyle/11152712
confidence: medium
basis: news_aggregation
period: statistics as of March 2026; Storm Media report of 2026-07-30
caveat: media citing Bureau statistics; the same report's "pension king drawing more than NT$52,000 a month" describes an individual case, not an average (see §Caveats)

F-012: CNA report of 2026-07-17: on appeal in the labor fund stock manipulation case, former Ministry of Labor section chief Yu Nai-wen and nine others received terms of 1 year 7 months to 8 years 4 months, 7 of them suspended, with 2 others still acquitted; the ruling may be appealed

Per the appellate press materials, Yu, then head of the domestic investment section of the Bureau of Labor Funds, bought 486 lots of Far Eastern Department Stores shares through the Bureau's proprietary account in August and September 2020 and instructed fund managers and analysts at Fuh Hwa and Uni-President securities investment trust companies, which managed labor fund money under mandate, to produce false investment assessment reports and buy 4,564 lots and 3,950 lots respectively; after all shares were sold and settled, the Bureau of Labor Funds recorded valuation losses of more than NT$13.63 million and more than NT$6.29 million respectively (cna#1568412).

source: CNA #1568412
source_url: https://www.cna.com.tw/news/asoc/202607170131.aspx
confidence: high
basis: news_aggregation
period: appellate ruling of 2026-07-17; the conduct period as reported runs from June 2019 to November 2020
caveat: the ruling may be appealed and is not final; ROC years 108 and 109 in the original are converted here; this card quotes only reported facts and sentences and offers no legal analysis (see §Caveats)

F-013: CNA report of 2026-07-17: Yu Nai-wen was sentenced to 9 years at first instance for offenses including violation of the Securities and Exchange Act, changed to 8 years 4 months on appeal; the Taipei District Court at first instance found criminal proceeds relating to the Pau Jar Group totaling NT$538,796,150

The same report states that the Taipei District Court calculated that Far Eastern Department Stores' share price rose from a close of NT$18.15 on June 19, 2019 to a close of NT$24.95 on November 27, 2020, a rise of 37.47% over that period, above the 6.25% of the trade and retail sector index and the 28.69% of the overall exchange market over that period (cna#1568015).

source: CNA #1568015
source_url: https://www.cna.com.tw/news/asoc/202607170068.aspx
confidence: high
basis: news_aggregation
period: report of 2026-07-17; share price comparison from the close of 2019-06-19 to the close of 2020-11-27
caveat: the criminal proceeds figure differs in basis from the NT$508,781,945 in the High Court press materials carried by cna#1568412, belonging to a different instance and scope of finding; this card neither combines nor adjudicates them (see §Caveats)

F-014: Taiwan's labor funds overall returned NT$2.1252 trillion for January–May 2026, a return rate of 27.7%; the ten-year average return rate is 12.24% and the five-year 15.94%, and the Bureau of Labor Funds forecast June 2026 to be flat

The CNA report of 2026-07-01 states May's single-month return was NT$586 billion, and that January, February, April and May 2026 were the four largest single-month returns in the labor funds' history (April's NT$1.1047 trillion the largest, then May's NT$586 billion, February's NT$418.4 billion and January's NT$394.6 billion). Deputy Director-General Liu Li-ju said the Taiwan stock market rose 54.45% from January to May 2026, twice the 25.70% rise for the whole of 2025 (cna#1283385).

source: CNA #1283385
source_url: https://www.cna.com.tw/news/ahel/202607010219.aspx
confidence: high
basis: news_aggregation
period: statistics through 2026-05-31; published by the Bureau of Labor Funds, Ministry of Labor, on 2026-07-01
caveat: returns are point-in-time statistics that move with markets; the flat June is a forecast and an outlook rather than a settled result; this figure is an institution-side investment result and is not comparable with Japanese individual survey percentages (see §Caveats)

F-015: Freeasy survey of 1,000 people (conducted June 30, 2026; men and women aged 15 and over nationwide): among investors, "no particular regret" led at 28.7%, and the leading specific regret was "started too late" at 19.5%

In the same survey, among those who would start or resume investing if their anxieties or barriers were resolved, what they most wanted from financial institutions was "being able to start with small amounts" at 45.1%, followed by "clear information for beginners" at 41.7%; the second and third specific regrets were "sold too early and missed the gains" at 18.0% and "was late to cut losses" at 17.5% (prtimes#1749426).

source: PRTIMES #1749426
source_url: https://prtimes.jp/main/html/rd/p/000000226.000064613.html
confidence: high
basis: official_statement
period: fieldwork 2026-06-30; released 2026-07-30
caveat: a company-run survey issued as a corporate release reviewed by the platform; the sample is 1,000 panel members of a self-service survey tool (equal cells by gender and age band) and is not nationally representative; the operator and question wording differ from the target card's 6,000-person survey, so figures are not combined (see §Caveats)

F-016: The background section of the same release cites Financial Services Agency data: NISA accounts stood at 28.26 million as of end-December 2025, a little over 20% of the total population, with cumulative purchases of 71 trillion yen (the release states this is quoted from Nikkei MJ of February 18, 2026)

This is a background citation within a corporate release, used to explain the context of rising interest in household asset formation in Japan (prtimes#1749426).

source: PRTIMES #1749426
source_url: https://prtimes.jp/main/html/rd/p/000000226.000064613.html
confidence: medium
basis: official_statement
period: statistics as of end-December 2025; released 2026-07-30
caveat: a second-hand citation within a corporate release (quoted from Nikkei MJ of 2026-02-18), not the Financial Services Agency's original statistics (see §Caveats)

F-017: Property Agent company-run survey of 545 people: 62.0% of those who consult AI had executed an investment on the AI answer alone without checking it, and 85.6% of that group (base 111) experienced losses or regret versus 19.1% of those who had not (base 68) — a gap of about 4.5x

The same survey reports information sources ever consulted for investment decisions as professionals 40.4%, YouTube videos 37.6%, newspapers and news sites 35.0%, social media 34.5%, generative AI 32.8% and family, friends and acquaintances 18.9%; among respondents in their 20s, generative AI led at 46.8%, ahead of professionals (prtimes#1765567).

source: PRTIMES #1765567
source_url: https://prtimes.jp/main/html/rd/p/000000381.000130175.html
confidence: high
basis: official_statement
period: released 2026-07-31
caveat: a company-run survey issued as a corporate release reviewed by the platform; the sample is limited to 545 company employees, directors, public servants and business owners who currently invest and have received a bonus, and is not nationally representative; this "about 4.5x" is a gap in experienced losses or regret and is not the same indicator as the target card's "about 4.4x" gap in intention (see §Caveats)

F-018: Other gauges from the same 545-person survey: 80.5% of generative AI users trust AI answers, 70.5% sometimes use them without checking, 43.1% prioritize AI over their own view in a conflict (above the 32.2% who prioritize themselves), 91.1% have adopted an AI answer as-is in some investment decision, and 70.7% want to keep using AI

The same survey also reports that 77.2% had put bonus money into investment, up 18.1 percentage points from the 59.1% in the company's previous survey of 550 people conducted on the same conditions in August 2025; the leading reason for consulting AI was "easy and gives an answer immediately" at 64.2%, while "I think the answers are accurate" was lowest at 10.6% (prtimes#1765567).

source: PRTIMES #1765567
source_url: https://prtimes.jp/main/html/rd/p/000000381.000130175.html
confidence: high
basis: official_statement
period: this survey released 2026-07-31; the previous survey was conducted 2025-08-19 to 2025-08-20 (550 people)
caveat: a company-run survey; "91.1%" is stated in the release as the inverse of the 8.9% who said they had never adopted an answer as-is; the two waves share conditions but are different samples, and this card quotes the release's own comparison as stated (see §Caveats)

F-019: Finfinity Inc.'s analysis of more than 1,000 consultations in its first year: the top three most-prioritized themes — life planning, asset management (NISA, iDeCo and similar) and preparing education funds — together account for about 67%, and 55.7% of consultees are in their 30s

The same release states that people in their 30s and 40s together account for about 74% and those in their 20s to 40s for nearly 90%, while 63.7% are married with children. The release describes the top three combined as "about 70%" in its lead and as "about 67%" in the body (prtimes#1749599).

source: PRTIMES #1749599
source_url: https://prtimes.jp/main/html/rd/p/000000011.000177318.html
confidence: medium
basis: official_statement
period: consultations accumulated over one year from launch; released 2026-07-30
caveat: the company's own business data, neither a market survey nor a nationally representative sample; the same release carries both "about 70%" and "about 67%," and this card quotes both as stated without adjudicating (see §Caveats)

F-020: As stated in the Japanese Financial Services Agency's press release of July 29, 2026: within the 2026 (Reiwa 8) amendments to the Financial Instruments and Exchange Act and related laws, the provisions raising penalties for unregistered business and adding criminal investigation powers for the Securities and Exchange Surveillance Commission take effect on August 12, 2026, with the related cabinet order decided on July 24, 2026 and promulgated on July 29, 2026

The same item also states that, because the order falls under the relevant provisions of the Administrative Procedure Act, no public comment procedure was carried out (fsa#1745443).

source: FSA #1745443
source_url: https://www.fsa.go.jp/news/r8/shouken/20260729/20260729.html
confidence: medium
basis: official_statement
period: cabinet order decided 2026-07-24 and promulgated 2026-07-29; the provisions take effect 2026-08-12
caveat: the content of that item as stated on the Financial Services Agency's press release page, not the full text of the original cabinet order itself; used only for institutional context and risk signalling, never as official backing for a principal conclusion (see §Caveats)

F-021: As stated in a corporate seminar announcement release on PR TIMES dated 2026-07-16: an amendment to the Defined Contribution Pension Act is scheduled for December 2026

NTT DATA ABIC Corporation's list of August 2026 web seminars for financial institutions includes one centered on "the content of the Defined Contribution Pension Act amendment scheduled for December 2026" (prtimes#1543126).

source: PRTIMES #1543126
source_url: https://prtimes.jp/main/html/rd/p/000000157.000115257.html
confidence: low
basis: official_statement
period: released 2026-07-16; the stated amendment timing is December 2026
caveat: a scheduled date as stated in a corporate seminar announcement, not an official confirmation notice; used only as context (see §Caveats)

J-Units

J-001: Taiwan and Japan each moved one step in the same month, but the positions are orthogonal — Taiwan moved the "entrance and its enforcement" (from 2026-08-01 employers may not refuse voluntary contributions and face surcharges if they do, with coverage widened to farmers in labor or national pension insurance), while Japan moved the "quality of the intermediary" (generative AI is now the fifth-ranked information source for investment decisions and first among people in their 20s, with 85.6% of those who acted on AI without checking reporting losses or regret)

confidence: medium
basis: news_aggregation

J-002: The target card's dichotomy of "central system investment versus individual choice" needs one REVISE: central investment concentrates investment performance (27.7% for January–May 2026) and governance risk (10 convicted on appeal in the manipulation case) at the same time, while an individual's eventual entitlement is set by the contribution rate (6% frozen for more than 20 years), wage level and years of service rather than by a single year's return — and Taiwan's return rate and Japan's behavioral survey percentages remain non-comparable

confidence: medium
basis: news_aggregation

J-003: The two multiples must remain permanently separate: the target card's about 4.4x is a gap in the "intention rate" by presence or absence of investment experience (6,000-person survey), while this card's about 4.5x is a gap in "losses or regret already experienced" by whether the person verified the answer (545-person survey, bases of 111 versus 68); describing them as movement in one indicator does not hold

confidence: high
basis: official_statement

P-Units

P-001: The actual number of notices issued and penalties imposed under the surcharge mechanism that began on August 1, 2026 — track subsequent Bureau of Labor Insurance statistics to test whether the "employer blocks voluntary contributions" bottleneck has really opened ### P-002: Whether the statutory minimum employer contribution rate moves from 6% (legislators' bills at 8% and 9%, labor groups at 12%) and when the Ministry of Labor's rough estimates become a formal proposal — track subsequent Legislative Yuan and Ministry announcements ### P-003: Whether "AI-dependent investing" appears in nationally representative surveys or supervisory statistics in Japan — this card cites a single company-run survey (545 respondents) and independent sources are needed; also track the actual content of the Defined Contribution Pension Act amendment scheduled for December 2026

Three Perspectives on the Same Event / 同事件・三視角 / 同一イベント・三つの視点

Internal citation chain

Published ANK-Docs cited by this article:

Extends | ANK-2026-07-03-004, "Taiwan-Japan Retirement 'Point of Contact With the Market' Comparison: a 6,000-Person Japanese Survey Found Only About 15% of People in Their 40s and 50s Intend to Put Retirement Lump Sums Into Investment, a Gap of About 4.4x by Investment Experience; Taiwan's Labor Funds Are Invested Centrally by the System, With a 27.7% Return for January–May 2026, and the Executive Yuan Approved Voluntary Contributions for Old-Scheme-Only Workers" (that card states a publication date of 2026-07-03) → this card is the deeper next version of the same subject: the target card stopped at "approved by the Executive Yuan, targeted for end-July implementation" and at "in Japan the individual's contact point is decided by investment experience," while this card advances the Taiwan side to implementation and enforcement power and the Japan side to "the intermediary has changed to generative AI," and pins down explicitly that the two cards' two multiples must not be mixed.
ANK-2026-07-18-006, "Voluntary Pension Contributions for Old-Scheme-Only Workers Took Effect on 2026-07-17 (About 115,000 Covered); Appellate Ruling in the Labor Fund Stock Manipulation Case the Same Day" (published 2026-07-20) → this card carries forward that card's record of the effective date and headcount to the enforcement power of the surcharge from 2026-08-01 and the nationwide implementation of the five amendments, and adds the sentences and the Bureau of Labor Funds' valuation losses from the appellate ruling.
ANK-2026-07-01-001, "The National Retirement Ledger Under an AI Bull Market: Labor Funds Returned a Record NT$2.1252 Trillion for January–May 2026 (a 27.7% Return Rate) and the Public Service Pension Fund NT$313.232 Billion (24.29%)" (2026-07-01) → provides the full breakdown behind this card's F-014 institution-side performance; this card only refers back and does not restate its sub-fund detail.
ANK-2026-07-03-007, "The Taiwan-Japan Retail Investing 'Temperature Gap': Japan's NISA Accounts Exceed 28 Million" (published 2026-07-02) → provides context alongside this card's F-016 NISA account scale (this card's figure is a second-hand citation from Nikkei MJ within a corporate release; each card attributes its own source and the figures are not combined).
ANK-2026-06-16-001, "The Great Retail Money Migration: Taiwan's Active ETFs Against Three Years of Structural Fissures in Japan's NISA 'From Savings to Investment'" (published 2026-06-25) → provides the longer-run structural background for this card's Japan-side individual contact point.

Sources

1. [CNA #1779735] CNA (reporter Wu Hsin-yun), "Employers Refusing Workers' Voluntary Pension Contributions Face Late-Payment Surcharges From August," 2026-08-01. https://www.cna.com.tw/news/ahel/202608010103.aspx 2. [PRTIMES #1765567] Property Agent Inc. (Migaro Holdings group), "Survey of 545 Experienced Investors: 'Investing at AI's Bidding' Emerges," 2026-07-31. https://prtimes.jp/main/html/rd/p/000000381.000130175.html 3. [STORM #1766274] Storm Media, "Attention, Older Workers Aged 60: Major Changes to Labor Pension Claim Rules — Taiwan's August Rule Changes at a Glance," 2026-07-31. https://www.storm.mg/lifestyle/11153034 4. [PRTIMES #1749426] i-Bridge Inc. (Freeasy), "A 1,000-Person Survey on the Realities of Investing and Asset Management," 2026-07-30. https://prtimes.jp/main/html/rd/p/000000226.000064613.html 5. [PRTIMES #1749599] Finfinity Inc., "Preparing for Education Costs and Retirement Funds: People in Their 30s Consult Most," 2026-07-30. https://prtimes.jp/main/html/rd/p/000000011.000177318.html 6. [STORM #1750606] Storm Media, "The Ministry of Labor Pays Out Today: Labor Insurance Pensions Average About NT$19,000 per Person — 12 Benefit Payment Dates at a Glance," 2026-07-30. https://www.storm.mg/lifestyle/11152712 7. [FSA #1745443] Financial Services Agency of Japan, "Promulgation of the Cabinet Order Relating to the 2026 (Reiwa 8) Amendments to the Financial Instruments and Exchange Act and Related Laws (Effective 20 Days After Promulgation)," 2026-07-29. https://www.fsa.go.jp/news/r8/shouken/20260729/20260729.html 8. [CNA #1723918] CNA (reporter Wu Hsin-yun), "Bureau of Labor Insurance Introduces Mobile Phone Authentication for Online Pension Estimates and Claims," 2026-07-28. https://www.cna.com.tw/news/ahel/202607280196.aspx 9. [STORM #1724024] Storm Media, "Major Ministry of Labor Amendments Take Effect: Five Changes to the New Labor Pension Scheme Applied Nationwide From August 1," 2026-07-28. https://www.storm.mg/lifestyle/11152226 10. [CNYES #1723739] Anue (written by the financial advisory team of Yung Cheng Asset Management), "The New Labor Pension Scheme Upgrades Again: a Reconsideration Window for Monthly Payments in August — Think Twice Before Contributing 6%," 2026-07-28. https://news.cnyes.com/news/id/6548369 11. [CNA #1721353] CNA (reporter Wu Hsin-yun), "Farmer Retirement Savings Coverage Widened: Farmers in Labor or National Pension Insurance May Contribute From August," 2026-07-28. https://www.cna.com.tw/news/ahel/202607280157.aspx 12. [STORM #1664683] Storm Media, "Employer Contributions of 6% Under the New Labor Pension Scheme Unchanged for More Than 20 Years; Labor Groups Call for an Increase and the Ministry of Labor Responds," 2026-07-23. https://www.storm.mg/article/11151357 13. [CNA #1658827] CNA (reporter Tseng Yi-ning), "Labor Insurance Revenue and Expenditure Cross Over; Ministry of Labor: Employer Contribution Rate Under Rough Calculation," 2026-07-23. https://www.cna.com.tw/news/aipl/202607230194.aspx 14. [STORM #1646661] Storm Media, "Do 60% of Workers Have Less Than NT$1 Million in Their Pension Accounts? Labor Groups on the Progress of Raising the 6% Employer Contribution Rate: 7.78 Million People's Nest Eggs at Risk," 2026-07-22. https://www.storm.mg/article/11151037 15. [CNA #1587714] CNA (reporter Wu Hsin-yun), "Voluntary Pension Contributions Opened to Old-Scheme-Only Workers: a Q&A," 2026-07-18. https://www.cna.com.tw/news/ahel/202607180082.aspx 16. [CNA #1568412] CNA (reporter Hsieh Chun-lin), "Labor Fund Stock Manipulation Case: Appellate Court Convicts Yu Nai-wen and Nine Others, Seven Suspended," 2026-07-17. https://www.cna.com.tw/news/asoc/202607170131.aspx 17. [CNA #1568015] CNA (reporter Hsieh Chun-lin), "Labor Fund Stock Manipulation Case: Yu Nai-wen's Sentence for Benefiting Pau Jar Changed to 8 Years 4 Months on Appeal," 2026-07-17. https://www.cna.com.tw/news/asoc/202607170068.aspx 18. [PRTIMES #1543126] NTT DATA ABIC Corporation, "August 2026 List of Web Seminars for Financial Institutions, Including 'The Defined Contribution Pension Act Amendment and DC Trends'," 2026-07-16. https://prtimes.jp/main/html/rd/p/000000157.000115257.html 19. [CNA #1283385] CNA (reporter Wu Hsin-yun), "Labor Funds Earned More Than NT$2 Trillion in January–May; June Forecast to Be Flat," 2026-07-01. https://www.cna.com.tw/news/ahel/202607010219.aspx

Internal citations (published ANK cards): ANK-2026-07-03-004 (extended from), ANK-2026-07-18-006, ANK-2026-07-01-001, ANK-2026-07-03-007, ANK-2026-06-16-001.

> Anti-fabrication statement: this card has zero verified official anchors (official_count_verified=0). All Taiwan figures are as stated in same-day media reports and are not figures verified by the competent authorities (for the scope of verification see §Caveats); all Japanese survey figures are as stated in company-run survey releases and are not nationally representative samples; the FSA cabinet order and the DC law amendment timing are only as stated in the item and the announcement release, and are not the original documents themselves. This card generates no figure absent from the sources, computes no ratio or total, and never merges figures across sources to manufacture a comparison. Taiwan's institution-side return rate and Japan's individual survey percentages are kept separate throughout, with no superiority comparison. No Wikidata Q identifiers are attached (no registry verification was performed for this piece).

FAQ

Q: From August 1, 2026, can an employer in Taiwan refuse a worker's voluntary pension contribution?

Huang Wei-chen, Director-General of the Ministry of Labor's Department of Labor Welfare and Retirement, explained that a worker must still express the intention to the employer first; under the new scheme the employer contributes at least 6% of monthly wages and the worker may add up to a further 6% (cna#1779735). This is as stated in the CNA report (see §Caveats).

Q: When does the 30-day reconsideration window for a monthly pension start?

What the three reports agree on is that the window applies to workers who chose monthly payments, that a switch to a lump sum is possible, and that per Storm Media the change may be made only once (cna#1779735, storm#1766274, cnyes#1723739).

Q: Does Taiwan's 27.7% labor fund return for January–May 2026 mean workers' pensions are sufficient?

Other structural signals during this card's observation window (2026-07-01 to 2026-08-01) include the statutory minimum employer contribution rate having stood at 6% since 2005 without adjustment for more than 20 years as of the July 2026 reports, and an academic's statement at the public hearing of 2026-07-23 that per Bureau of Labor Insurance statistics labor insurance premium outlays exceeded income in 2025 (cna#1283385, storm#1646661, storm#1664683, cna#1658827).

Q: Is Japan's "about 4.5x" here the same thing as the target card's "about 4.4x"?

The two surveys have entirely different operating companies, population conditions and question wording, and both are company-run surveys rather than nationally representative samples (prtimes#1765567). Any statement that strings the two ratios of the target card (published 2026-07-03) and this card (published 2026-08-02) together as "Japan's investment gap widened from 4.4x to 4.5x" does not hold.

Q: What moved on Japan's institutional side during this month?

Both are used only for institutional context and risk signalling, not as official endorsement (fsa#1745443, prtimes#1543126; for the scope of verification see §Caveats). Where Taiwan moved the enforcement power at the entrance over the same period, the position Japan moved sits closer to market intermediation and supervision — this juxtaposition is this card's editorial inference and is not a sentence from any source.

📊 引用級事實單元（F-Units）

From August 1, 2026, Taiwanese employers may not refuse a worker's voluntary pension contribution; if they still fail to act after a Bureau of Labor Insurance notice, a surcharge accrues at 3% of the amount due per day overdue, capped at one times that amount

F-001 · Confidence: high · Basis: news_aggregation CNA #1779735 Grade C · 媒體報導（非企業審查） in force from 2026-08-01; CNA report of 2026-08-01

Storm Media report of 2026-07-28: five amendments to the new labor pension scheme take nationwide uniform effect from August 1, 2026, with an estimated impact on about 13.1 million workers holding individual labor pension accounts

F-002 · Confidence: medium · Basis: news_aggregation STORM #1724024 Grade C · 媒體報導（非企業審查） effective 2026-08-01; Storm Media report of 2026-07-28

Storm Media report of 2026-07-31: workers with at least 15 years of contributions may apply to switch to a lump sum within 30 days of the first monthly pension payment being credited, and the ten-year claim period for minor survivors now runs from "the date of reaching the age of 20"

F-003 · Confidence: medium · Basis: news_aggregation STORM #1766274 Grade C · 媒體報導（非企業審查） from 2026-08-01; Storm Media report of 2026-07-31

Anue report of 2026-07-28: the 30-day reconsideration window for monthly pensions starts August 1, 2026 (that report words it as 30 days "before approval" to switch to a lump sum), and voluntary contributions by old-scheme-only workers began on July 17, 2026

F-004 · Confidence: medium · Basis: news_aggregation CNYES #1723739 Grade C · 媒體報導（非企業審查） from 2026-07-17 (old-scheme voluntary contributions) and from 2026-08-01 (reconsideration window); Anue report of 2026-07-28

CNA Q&A of 2026-07-18: old-scheme-only workers are those hired before July 1, 2005 who chose to stay under the old scheme, currently about 115,000 people; voluntary contributions are excluded from that year's taxable wage income and the account carries a minimum return guarantee of at least the two-year time deposit rate

F-005 · Confidence: high · Basis: news_aggregation CNA #1587714 Grade C · 媒體報導（非企業審查） CNA report of 2026-07-18; headcount as stated in that report at that time

CNA report of 2026-07-28: the government's share of farmer retirement savings contributions was raised to 60% (applicable from January 2026), and from August 2026 coverage widened to actual farmers enrolled in labor or national pension insurance; 88,000 people currently contribute, most at a monthly rate of 10%

F-006 · Confidence: high · Basis: news_aggregation CNA #1721353 Grade C · 媒體報導（非企業審查） coverage widened from August 2026; CNA report of 2026-07-28 (briefing by Wen Hsiu-chu, head of the Bureau's farmer insurance division, at a Ministry of Labor regular meeting)

CNA report of 2026-07-28: the Bureau of Labor Insurance's e-services introduced "mobile phone authentication," letting workers aged 60 or over under the new scheme estimate and claim their pension online around the clock

F-007 · Confidence: high · Basis: news_aggregation CNA #1723918 Grade C · 媒體報導（非企業審查） CNA report of 2026-07-28 (Ministry of Labor regular meeting)

May First Action Alliance release of 2026-07-22: nearly 7.78 million workers are covered by the new labor pension scheme, and at the current 6% contribution rate "as many as 60% of workers accumulate less than NT$1 million of principal in their individual accounts after 25 years of work"

F-008 · Confidence: medium · Basis: news_aggregation STORM #1646661 Grade C · 媒體報導（非企業審查） report of 2026-07-22 (May First Action Alliance release)

Storm Media report of 2026-07-23: the statutory minimum employer contribution rate under the Labor Pension Act has been 6% since the act took effect in 2005 and, as of the July 2026 report, had gone more than 20 years without adjustment; legislators' bills propose 8% or 9% and labor groups seek a phased increase to 12%

F-009 · Confidence: medium · Basis: news_aggregation STORM #1664683 Grade C · 媒體報導（非企業審查） report of 2026-07-23 (Legislative Yuan Social Welfare and Environmental Hygiene Committee public hearing)

CNA report of 2026-07-23: Deputy Minister of Labor Huang Ling-na said the Ministry has done rough estimates and internal discussion on raising the employer contribution rate and that an increase would affect more than 600,000 business units and about 8 million covered workers; academic Hao Chung-jen said that per Bureau of Labor Insurance statistics, labor insurance premium outlays exceeded income in 2025, a "death cross"

F-010 · Confidence: medium · Basis: news_aggregation CNA #1658827 Grade C · 媒體報導（非企業審查） public hearing of 2026-07-23; the "death cross" refers to 2025 as reported (the original says "last year")

Storm Media on 2026-07-30, citing the Bureau of Labor Insurance's latest statistics: as of March 2026, 2.102 million workers were drawing labor insurance pensions, with monthly disbursements totaling NT$40.16 billion and an average of NT$19,105 per person per month

F-011 · Confidence: medium · Basis: news_aggregation STORM #1750606 Grade C · 媒體報導（非企業審查） statistics as of March 2026; Storm Media report of 2026-07-30

CNA report of 2026-07-17: on appeal in the labor fund stock manipulation case, former Ministry of Labor section chief Yu Nai-wen and nine others received terms of 1 year 7 months to 8 years 4 months, 7 of them suspended, with 2 others still acquitted; the ruling may be appealed

F-012 · Confidence: high · Basis: news_aggregation CNA #1568412 Grade C · 媒體報導（非企業審查） appellate ruling of 2026-07-17; the conduct period as reported runs from June 2019 to November 2020

CNA report of 2026-07-17: Yu Nai-wen was sentenced to 9 years at first instance for offenses including violation of the Securities and Exchange Act, changed to 8 years 4 months on appeal; the Taipei District Court at first instance found criminal proceeds relating to the Pau Jar Group totaling NT$538,796,150

F-013 · Confidence: high · Basis: news_aggregation CNA #1568015 Grade C · 媒體報導（非企業審查） report of 2026-07-17; share price comparison from the close of 2019-06-19 to the close of 2020-11-27

Taiwan's labor funds overall returned NT$2.1252 trillion for January–May 2026, a return rate of 27.7%; the ten-year average return rate is 12.24% and the five-year 15.94%, and the Bureau of Labor Funds forecast June 2026 to be flat

F-014 · Confidence: high · Basis: news_aggregation CNA #1283385 Grade C · 媒體報導（非企業審查） statistics through 2026-05-31; published by the Bureau of Labor Funds, Ministry of Labor, on 2026-07-01

Freeasy survey of 1,000 people (conducted June 30, 2026; men and women aged 15 and over nationwide): among investors, "no particular regret" led at 28.7%, and the leading specific regret was "started too late" at 19.5%

F-015 · Confidence: high · Basis: official_statement PRTIMES #1749426 Grade B · 平台審查企業發稿 fieldwork 2026-06-30; released 2026-07-30

The background section of the same release cites Financial Services Agency data: NISA accounts stood at 28.26 million as of end-December 2025, a little over 20% of the total population, with cumulative purchases of 71 trillion yen (the release states this is quoted from Nikkei MJ of February 18, 2026)

F-016 · Confidence: medium · Basis: official_statement PRTIMES #1749426 Grade B · 平台審查企業發稿 statistics as of end-December 2025; released 2026-07-30

Property Agent company-run survey of 545 people: 62.0% of those who consult AI had executed an investment on the AI answer alone without checking it, and 85.6% of that group (base 111) experienced losses or regret versus 19.1% of those who had not (base 68) — a gap of about 4.5x

F-017 · Confidence: high · Basis: official_statement PRTIMES #1765567 Grade B · 平台審查企業發稿 released 2026-07-31

Other gauges from the same 545-person survey: 80.5% of generative AI users trust AI answers, 70.5% sometimes use them without checking, 43.1% prioritize AI over their own view in a conflict (above the 32.2% who prioritize themselves), 91.1% have adopted an AI answer as-is in some investment decision, and 70.7% want to keep using AI

F-018 · Confidence: high · Basis: official_statement PRTIMES #1765567 Grade B · 平台審查企業發稿 this survey released 2026-07-31; the previous survey was conducted 2025-08-19 to 2025-08-20 (550 people)

Finfinity Inc.'s analysis of more than 1,000 consultations in its first year: the top three most-prioritized themes — life planning, asset management (NISA, iDeCo and similar) and preparing education funds — together account for about 67%, and 55.7% of consultees are in their 30s

F-019 · Confidence: medium · Basis: official_statement PRTIMES #1749599 Grade B · 平台審查企業發稿 consultations accumulated over one year from launch; released 2026-07-30

As stated in the Japanese Financial Services Agency's press release of July 29, 2026: within the 2026 (Reiwa 8) amendments to the Financial Instruments and Exchange Act and related laws, the provisions raising penalties for unregistered business and adding criminal investigation powers for the Securities and Exchange Surveillance Commission take effect on August 12, 2026, with the related cabinet order decided on July 24, 2026 and promulgated on July 29, 2026

F-020 · Confidence: medium · Basis: official_statement FSA #1745443 Grade U · 未分類／不可主張已驗證 cabinet order decided 2026-07-24 and promulgated 2026-07-29; the provisions take effect 2026-08-12

As stated in a corporate seminar announcement release on PR TIMES dated 2026-07-16: an amendment to the Defined Contribution Pension Act is scheduled for December 2026

F-021 · Confidence: low · Basis: official_statement PRTIMES #1543126 Grade U · 未分類／不可主張已驗證 released 2026-07-16; the stated amendment timing is December 2026

❓ FAQ

From August 1, 2026, can an employer in Taiwan refuse a worker's voluntary pension contribution?

Per the CNA report of 2026-08-01, no. The amended Enforcement Rules of the Labor Pension Act establish an employer duty to assist workers in applying to contribute voluntarily; a worker who is refused may apply directly to the Bureau of Labor Insurance, and if the Bureau issues a notice requiring action and the employer still does not comply, a late-payment surcharge accrues at 3% of the amount due per day overdue, capped at one times the amount due. Huang Wei-chen, Director-General of the Ministry of Labor's Department of Labor Welfare and Retirement, explained that a worker must still express the intention to the employer first; under the new scheme the employer contributes at least 6% of monthly wages and the worker may add up to a further 6% (cna#1779735). This is as stated in the CNA report (see §Caveats).

When does the 30-day reconsideration window for a monthly pension start?

The three outlets word it differently and this card does not adjudicate: CNA on 2026-08-01 says "30 days after the first payment is credited," Storm Media on 2026-07-31 says "within 30 days of the first monthly pension payment being credited," and Anue on 2026-07-28 says "30 days after applying, before approval." Anyone exercising the right should rely on the Bureau of Labor Insurance's announcements. What the three reports agree on is that the window applies to workers who chose monthly payments, that a switch to a lump sum is possible, and that per Storm Media the change may be made only once (cna#1779735, storm#1766274, cnyes#1723739).

Does Taiwan's 27.7% labor fund return for January–May 2026 mean workers' pensions are sufficient?

That inference does not follow. The 27.7% (NT$2.1252 trillion in returns, statistics through 2026-05-31) is an institution-side period investment result, while the final balance of an individual account depends on the contribution rate, wages and years of service. Using the current 6% contribution rate, the May First Action Alliance release of 2026-07-22 estimates that as many as 60% of workers accumulate less than NT$1 million of principal after 25 years of work (a labor-group figure, not official statistics). Other structural signals during this card's observation window (2026-07-01 to 2026-08-01) include the statutory minimum employer contribution rate having stood at 6% since 2005 without adjustment for more than 20 years as of the July 2026 reports, and an academic's statement at the public hearing of 2026-07-23 that per Bureau of Labor Insurance statistics labor insurance premium outlays exceeded income in 2025 (cna#1283385, storm#1646661, storm#1664683, cna#1658827).

Is Japan's "about 4.5x" here the same thing as the target card's "about 4.4x"?

No, and neither may be inferred from or combined with the other. The target card's about 4.4x is the gap in the intention to put a retirement lump sum into investment by presence or absence of investment experience in a 6,000-person survey (25.1% versus 5.7%). This card's about 4.5x is the gap in losses or regret already experienced in a 545-person survey — 85.6% among those who executed on AI without checking versus 19.1% among those who did not (bases of 111 and 68 respectively). The two surveys have entirely different operating companies, population conditions and question wording, and both are company-run surveys rather than nationally representative samples (prtimes#1765567). Any statement that strings the two ratios of the target card (published 2026-07-03) and this card (published 2026-08-02) together as "Japan's investment gap widened from 4.4x to 4.5x" does not hold.

What moved on Japan's institutional side during this month?

As stated in the Financial Services Agency's press release item of 2026-07-29, within the 2026 (Reiwa 8) amendments to the Financial Instruments and Exchange Act and related laws, the provisions raising penalties for unregistered business and adding criminal investigation powers for the Securities and Exchange Surveillance Commission take effect on August 12, 2026, with the accompanying cabinet order decided on July 24, 2026 and promulgated on July 29, 2026. Separately, per a corporate seminar announcement release, an amendment to the Defined Contribution Pension Act is scheduled for December 2026. Both are used only for institutional context and risk signalling, not as official endorsement (fsa#1745443, prtimes#1543126; for the scope of verification see §Caveats). Where Taiwan moved the enforcement power at the entrance over the same period, the position Japan moved sits closer to market intermediation and supervision — this juxtaposition is this card's editorial inference and is not a sentence from any source. ---

🧠 編輯判斷（J-Units）

Taiwan and Japan each moved one step in the same month, but the positions are orthogonal — Taiwan moved the "entrance and its enforcement" (from 2026-08-01 employers may not refuse voluntary contributions and face surcharges if they do, with coverage widened to farmers in labor or national pension insurance), while Japan moved the "quality of the intermediary" (generative AI is now the fifth-ranked information source for investment decisions and first among people in their 20s, with 85.6% of those who acted on AI without checking reporting losses or regret)

Confidence: medium

The target card's dichotomy of "central system investment versus individual choice" needs one REVISE: central investment concentrates investment performance (27.7% for January–May 2026) and governance risk (10 convicted on appeal in the manipulation case) at the same time, while an individual's eventual entitlement is set by the contribution rate (6% frozen for more than 20 years), wage level and years of service rather than by a single year's return — and Taiwan's return rate and Japan's behavioral survey percentages remain non-comparable

Confidence: medium

The two multiples must remain permanently separate: the target card's about 4.4x is a gap in the "intention rate" by presence or absence of investment experience (6,000-person survey), while this card's about 4.5x is a gap in "losses or regret already experienced" by whether the person verified the answer (545-person survey, bases of 111 versus 68); describing them as movement in one indicator does not hold

Confidence: high

🔮 待驗證假設（P-Units）

The actual number of notices issued and penalties imposed under the surcharge mechanism that began on August 1, 2026 — track subsequent Bureau of Labor Insurance statistics to test whether the "employer blocks voluntary contributions" bottleneck has really opened

Status: open

Whether the statutory minimum employer contribution rate moves from 6% (legislators' bills at 8% and 9%, labor groups at 12%) and when the Ministry of Labor's rough estimates become a formal proposal — track subsequent Legislative Yuan and Ministry announcements

Status: open

Whether "AI-dependent investing" appears in nationally representative surveys or supervisory statistics in Japan — this card cites a single company-run survey (545 respondents) and independent sources are needed; also track the actual content of the Defined Contribution Pension Act amendment scheduled for December 2026

Status: open

TL;DR

Main text

1. Taiwan: the entrance went from "opened" to "may not be refused"

2. The other four items effective the same day: reconsideration window, survivor time limits, account protection, old-scheme contributions

3. Widening coverage: farmer retirement savings and online claims

4. CHALLENGE: the other side of 27.7%

5. Japan: the contact point has not disappeared, but the intermediary has changed

6. CONNECT: both places moved the system, but in different places

§Caveats (source limits, collected in one place)

Extension layers and typed actions

F-Units

F-001: From August 1, 2026, Taiwanese employers may not refuse a worker's voluntary pension contribution; if they still fail to act after a Bureau of Labor Insurance notice, a surcharge accrues at 3% of the amount due per day overdue, capped at one times that amount

F-002: Storm Media report of 2026-07-28: five amendments to the new labor pension scheme take nationwide uniform effect from August 1, 2026, with an estimated impact on about 13.1 million workers holding individual labor pension accounts

F-003: Storm Media report of 2026-07-31: workers with at least 15 years of contributions may apply to switch to a lump sum within 30 days of the first monthly pension payment being credited, and the ten-year claim period for minor survivors now runs from "the date of reaching the age of 20"

F-004: Anue report of 2026-07-28: the 30-day reconsideration window for monthly pensions starts August 1, 2026 (that report words it as 30 days "before approval" to switch to a lump sum), and voluntary contributions by old-scheme-only workers began on July 17, 2026

F-007: CNA report of 2026-07-28: the Bureau of Labor Insurance's e-services introduced "mobile phone authentication," letting workers aged 60 or over under the new scheme estimate and claim their pension online around the clock

F-008: May First Action Alliance release of 2026-07-22: nearly 7.78 million workers are covered by the new labor pension scheme, and at the current 6% contribution rate "as many as 60% of workers accumulate less than NT$1 million of principal in their individual accounts after 25 years of work"

F-011: Storm Media on 2026-07-30, citing the Bureau of Labor Insurance's latest statistics: as of March 2026, 2.102 million workers were drawing labor insurance pensions, with monthly disbursements totaling NT$40.16 billion and an average of NT$19,105 per person per month

F-012: CNA report of 2026-07-17: on appeal in the labor fund stock manipulation case, former Ministry of Labor section chief Yu Nai-wen and nine others received terms of 1 year 7 months to 8 years 4 months, 7 of them suspended, with 2 others still acquitted; the ruling may be appealed

F-014: Taiwan's labor funds overall returned NT$2.1252 trillion for January–May 2026, a return rate of 27.7%; the ten-year average return rate is 12.24% and the five-year 15.94%, and the Bureau of Labor Funds forecast June 2026 to be flat

F-015: Freeasy survey of 1,000 people (conducted June 30, 2026; men and women aged 15 and over nationwide): among investors, "no particular regret" led at 28.7%, and the leading specific regret was "started too late" at 19.5%

F-017: Property Agent company-run survey of 545 people: 62.0% of those who consult AI had executed an investment on the AI answer alone without checking it, and 85.6% of that group (base 111) experienced losses or regret versus 19.1% of those who had not (base 68) — a gap of about 4.5x

F-019: Finfinity Inc.'s analysis of more than 1,000 consultations in its first year: the top three most-prioritized themes — life planning, asset management (NISA, iDeCo and similar) and preparing education funds — together account for about 67%, and 55.7% of consultees are in their 30s

F-021: As stated in a corporate seminar announcement release on PR TIMES dated 2026-07-16: an amendment to the Defined Contribution Pension Act is scheduled for December 2026

J-Units

P-Units

Three Perspectives on the Same Event / 同事件・三視角 / 同一イベント・三つの視点

Internal citation chain

Sources

FAQ

Q: From August 1, 2026, can an employer in Taiwan refuse a worker's voluntary pension contribution?

Q: When does the 30-day reconsideration window for a monthly pension start?

Q: Does Taiwan's 27.7% labor fund return for January–May 2026 mean workers' pensions are sufficient?

Q: Is Japan's "about 4.5x" here the same thing as the target card's "about 4.4x"?

Q: What moved on Japan's institutional side during this month?

📊 引用級事實單元（F-Units）

❓ FAQ

🧠 編輯判斷（J-Units）

🔮 待驗證假設（P-Units）

Verification Record

Related Analyses