Same Year, Same Three Segments, Two Totals

Two tables in the same part of H.U. Group Holdings' annual securities report cover the same twelve months, split the company into the same three segments, and do not agree. One totals ¥265,097 million (US$1,618.4 million). The other totals ¥247,362 million (US$1,510.1 million), which is also consolidated revenue for the year ended 31 March 2026, the 76th term, filed on 15 June 2026. The difference between them is ¥17,735 million (US$108.3 million), about 7.2 per cent of the smaller figure (17,735 ÷ 247,362).

Segment by segment: laboratory testing and related services appears as ¥155,229 million (US$947.7 million) in one table and ¥157,297 million (US$960.3 million) in the other; in-vitro diagnostics as ¥82,564 million (US$504.1 million) and ¥60,735 million (US$370.8 million); healthcare-related services as ¥27,302 million (US$166.7 million) and ¥29,330 million (US$179.1 million). Two of the three differences are small, minus ¥2,068 million and minus ¥2,028 million (US$12.6 million and US$12.4 million). The third is ¥21,829 million (US$133.3 million). The three net to ¥17,733 million, two million yen (US$12,000) short of the difference between the totals, because the first table's three lines sum to ¥265,095 million against the ¥265,097 million printed.

One of the two sets is identifiable. The earnings release of 14 May 2026 prints the second set exactly, segment by segment, and states that segment revenue there means revenue from external customers. Those three lines add to consolidated revenue with nothing left over, which is what segment revenue looks like once intra-group traffic is out of it. What the other table measures we are not going to say. In the excerpt we read, its column headings did not come through attached to its numbers, and inferring a heading is not the same as reading one.

The two tables also disagree about which way the diagnostics year went, and the second one needs help to say so. The first prints its own year-on-year ratios: diagnostics at 97.2 per cent of the prior year, laboratory testing at 103.1, healthcare services at 99.5. The second prints one ratio only, 101.8 per cent, for the total. For its segments the direction comes from elsewhere. The annual report's narrative gives the movements in words and percentages — laboratory testing up 2.8 per cent, diagnostics up 0.4, healthcare services down 0.6 — and the prior-year amounts those changes are measured against are printed not in either table but in the May release: ¥153,014 million (US$934.2 million), ¥60,492 million (US$369.3 million) and ¥29,518 million (US$180.2 million). The arithmetic reproduces the narrative: 60,735 ÷ 60,492 is 1.004, 157,297 ÷ 153,014 is 1.028, 29,330 ÷ 29,518 is 0.994. Two of the three land within a rounding step of the first table's printed ratios. The third does not: 97.2 per cent against 100.4. The disagreement is not spread across the company. It sits in one segment, and it changes the sign.

Now the thing that did not happen. Those three prior-year revenue lines also appear in the annual report filed 16 June 2025, where they sit beside a laboratory operating loss of ¥4,638 million (US$28.3 million), a diagnostics profit of ¥11,345 million (US$69.3 million) and healthcare services at ¥1,777 million (US$10.8 million). All six figures appear again, unchanged, in the prior-year column of the earnings release eleven months later. Whatever moved between those two documents, these six did not.

The names have been stable longer than that. The integrated report dates the move to the present three segments, LTS, IVD and HS, to 2021. Its ten-year financial series carries one scope note, which reads in the original: 「2015-2017年度は海外検査事業（MLS）を除く」, excluding the overseas testing business from three early years (H.U. Group integrated report). That is the kind of note that tells a reader a boundary moved and that the series on either side of it is not one series. Beyond it, the excerpt of the series we read gives no further indication of a restatement, which is not the same as a positive statement that the last two years sit on one basis. For those two years, what we have is the six figures above, printed twice, eleven months apart, identical.

One boundary did move inside the year, and the arithmetic shows it moving. Healthcare-related services grew 8.9 per cent in the first quarter, ¥7,992 million (US$48.8 million) against ¥7,341 million (US$44.8 million), and shrank 0.6 per cent across the twelve months, ¥29,330 million against ¥29,518 million. The annual report attributes the full-year decline in part to ケアレックス becoming an equity-method affiliate. A company that moves from consolidation to the equity method stops contributing revenue and starts contributing a share of profit. The segment keeps its name and loses a set of lines.

The laboratory business is where the year's improvement in operating profit came from, and it repays being precise about what can be separated and what cannot. Revenue rose ¥4,283 million (US$26.1 million). The result moved from a loss of ¥4,638 million to a profit of ¥31 million (US$0.19 million) — a swing of ¥4,669 million (US$28.5 million), larger than the revenue increase, though the segment's own result at the end of it is still ¥31 million against consolidated operating profit of ¥4,780 million (US$29.2 million). Even if every incremental yen had carried a 100 per cent contribution margin, at least ¥386 million (US$2.4 million) of the improvement has to come from the pre-existing base. The company points to base-business revenue growth, an increase in marginal profit led by price adjustment, and improved testing operations centred on the fully operating H.U. Bioness Complex; elsewhere the report dates the あきる野 lab's full operation to the first quarter of the year just ended. The three arrive as description rather than as amounts, and on these disclosures volume, price and cost arrive as one number. The quarterly filing adds shape: a first-quarter loss of ¥800 million (US$4.9 million) implies ¥831 million (US$5.1 million) of profit across the remaining nine months, against a first-quarter loss of ¥1,636 million (US$10.0 million) and a further ¥3,002 million (US$18.3 million) of loss over the rest of the prior year.

Below the segments sits the adjustment. The three segment results sum to ¥10,840 million against a printed segment total of ¥10,841 million (US$66.2 million); the release states that it truncates below one million yen, and a one-million-yen difference is what that convention produces. Subtract the adjustment of ¥6,061 million (US$37.0 million), which holds inter-segment eliminations and unallocated corporate cost, and consolidated operating profit is ¥4,780 million against ¥2,640 million (US$16.1 million) a year earlier. The bridge is short enough to write out. The laboratory swing of ¥4,669 million, less ¥2,295 million (US$14.0 million) of decline in diagnostics profit and ¥18 million (US$0.11 million) in healthcare services, is ¥2,356 million (US$14.4 million); added to the prior year's segment total of ¥8,484 million (US$51.8 million) it gives ¥10,840 million again. The printed change in the segment total is ¥2,357 million (US$14.4 million). Take off ¥217 million (US$1.3 million), by which the adjustment deepened from ¥5,844 million (US$35.7 million), and ¥2,140 million (US$13.1 million) is what reaches the consolidated line. The company prints an 81.0 per cent increase. Dividing the two printed figures gives 81.1. Both sit inside the truncation.

Three profit lines went three ways in one year. Operating profit rose by about four-fifths. Ordinary profit fell 40.2 per cent to ¥2,834 million (US$17.3 million), which the company attributes to investment gains recorded in the prior year and not repeated. Net profit attributable to owners rose 147.1 per cent to ¥6,823 million (US$41.7 million), which it attributes to gains on sales of fixed assets and of shares in affiliates. The operating line and the bottom line are describing different events, and each is labelled with the event it describes.

The cash statement records the same year in a different measure. Operating cash flow was ¥21,565 million (US$131.7 million) against ¥21,964 million (US$134.1 million), a decline of ¥399 million (US$2.4 million). Investing moved from ¥15,958 million (US$97.4 million) of cash used to ¥11,339 million (US$69.2 million) provided. Financing used ¥26,393 million (US$161.1 million) against ¥5,298 million (US$32.3 million), roughly five times the prior year. Two years earlier the group printed free cash flow of ¥500 million (US$3.1 million); the mid-term plan's version of that measure keeps leases out of investing, so free cash flow arrives here in more than one construction and this year's is not something the two lines above let us rebuild. Cash and equivalents ended the year at ¥48,104 million (US$293.7 million).

Research spending is where the choice between the two revenue tables stops being abstract. Diagnostics research and development was ¥9,084 million (US$55.5 million). Over the ¥60,735 million that the May release and the annual report's second table both print, it is 15.0 per cent (9,084 ÷ 60,735). Over the ¥82,564 million the other table shows, it is 11.0 per cent (9,084 ÷ 82,564). The excerpt does not state the basis of that second figure, so the denominator is a choice the reader makes rather than one the page makes. A year earlier the same research line was ¥8,857 million (US$54.1 million), or 14.6 per cent of the prior year's ¥60,492 million, and it grew 2.6 per cent (9,084 ÷ 8,857). The mid-term plan says the diagnostics business invests 12 per cent of sales in research on a continuing basis, and that figure lies between the two ratios the annual report's own tables allow one to compute. Group research was ¥11,171 million (US$68.2 million), or 4.5 per cent of consolidated revenue, against a plan that speaks of holding a 5 per cent level; the laboratory segment's own research line is ¥466 million (US$2.8 million).

Capital spending fell by a third, to ¥7,608 million (US$46.4 million) from ¥11,515 million (US$70.3 million), a 33.9 per cent decline. The four lines — laboratory ¥3,870 million (US$23.6 million), diagnostics ¥1,964 million (US$12.0 million), healthcare ¥1,202 million (US$7.3 million), corporate ¥570 million (US$3.5 million) — sum to ¥7,606 million, two million short of the printed total. The reason for the difference is not stated in the filings; rounding at the million-yen mark is the mechanism that produces a gap of that size. The plan calls for roughly ¥40,000 million (US$244.2 million) of capital spending across five years. Year one used 19.0 per cent of it.

Depreciation was ¥21,139 million (US$129.1 million), 4.4 times operating profit, and goodwill amortisation ¥618 million (US$3.8 million). Under the definition the company prints — operating profit plus depreciation plus goodwill amortisation — the precise figures give ¥4,780 million + ¥21,139 million + ¥618 million = ¥26,537 million (US$162.0 million). The supplementary table states EBITDA in units of ¥100 million, four-rounded, and shows ¥26,500 million (US$161.8 million); built instead from that table's own rounded operating profit of ¥4,800 million (US$29.3 million), the same sum is ¥26,557 million (US$162.1 million), which sits on the other side of the ¥100 million step. Which of the two inputs was rounded, and when, is not something the table says. The plan's depreciation line falls from ¥13,000 million (US$79.4 million) for the year just ended to ¥8,500 million (US$51.9 million) for the year ending March 2028 and ¥7,000 million (US$42.7 million) for the year ending March 2030, printed under the discussion of the Complex reaching full operation.

The destination for the year ending March 2030 is printed in two forms. The May 2025 plan gives consolidated EBITDA of ¥30,500 million (US$186.2 million) at a 12.1 per cent margin and operating profit of ¥8,000 million (US$48.8 million) at 3.2 per cent; both imply a revenue base near ¥250,000 million (US$1,526.3 million), since 30,500 ÷ 0.121 is 252,066 and 8,000 ÷ 0.032 is 250,000. The annual report and the integrated report give, for the same year, an EBITDA margin of 16 per cent or more and an operating margin of 11 per cent or more. The plan pairs its margins with yen amounts; the other two state theirs as floors. ROE of 13 per cent or more and cumulative operating cash flow of ¥150,000 million (US$915.8 million) or more appear in all three. The prior plan's realised figures were a 9.6 per cent EBITDA margin and a 1.1 per cent operating margin.

For the year now running the company forecasts revenue of ¥256,000 million (US$1,562.9 million), up 3.5 per cent; operating profit of ¥9,000 million (US$54.9 million), up 88.3 per cent; and net profit of ¥5,000 million (US$30.5 million), down 26.7 per cent. Its two return measures move in opposite directions in its own table, ROE from 5.0 to 3.7 per cent and ROIC from 1.5 to 3.0. Forecast earnings per share of ¥90.11 (US$0.5501) against an unchanged dividend of ¥125.00 (US$0.7631) gives the 138.7 per cent payout printed alongside (125 ÷ 90.11).

Treasury shares appear twice with different counts, and the difference is documented rather than left to inference. The annual report puts holdings at 493,922 shares. The earnings release puts period-end treasury at 2,006,640 and says the figure includes shares held by the share-compensation trusts. The difference is 1,512,718 shares, which is exactly the trust holding the release discloses elsewhere. Same concept, two boundaries, both stated on the page where they are used.

The payout ratio behaves the same way. Dividends declared for the year total ¥7,138 million (US$43.6 million) against net profit of ¥6,823 million, which is 104.6 per cent. The company prints 102.9. Remove the ¥119 million (US$0.73 million) of dividends paid on trust-held shares and (7,138 − 119) ÷ 6,823 is 102.9 per cent; per share, ¥125.00 of dividend over ¥121.52 (US$0.7419) of earnings gives the same 102.9. The trust is inside one number and outside the other, and each is right on its own terms.

The buyback authorised in May 2025 carried two limits: 2,500,000 shares and ¥5,000,000,000 (US$30.5 million). It used 1,350,400 shares, 54.0 per cent of the share limit, and ¥4,999,762,717, or 99.995 per cent of the yen limit. What went unused was ¥237,283 (US$1,449). The yen cap bound; the share cap did not. A further ¥5,000 million (US$30.5 million) was resolved in May 2026, against a five-year framework of ¥20,000 million (US$122.1 million) or more of buybacks alongside a progressive dividend.

One more figure has to be assembled rather than read. Our own count of the shareholder table gives foreign ownership of 53.84 per cent as at 31 March 2026: 308,859 units held by foreign corporations plus 64 units held by foreign individuals, over 573,794 units in all. The denominator checks against the share count — 57,492,322 shares issued, less 112,922 held in odd lots, is 57,379,400, which is 573,794 units of one hundred. The form prints the two components as 53.83 and 0.01 per cent and has no total column, so the addition belongs to the reader. Across the 3,167 companies we measure the same way, the median is 7.31 per cent.

On conversion: every yen figure in this piece is converted at ¥163.8 = US$1, the rate on 24 July 2026, applied uniformly to all periods, including years ended in 2024, 2025 and 2026 and to forecast and plan figures. No period-specific or average rates are used, so dollar figures across years reflect only the yen amounts, not currency movement. Dollar amounts keep the significant figures of the yen amounts as printed by the company; ratios, percentages, multiples and share and unit counts are left unconverted. Worked example: ¥247,362 million ÷ 163.8 = US$1,510.1 million.