Chief Human Resources Officer / Japan
Decisions that cost tenders, payroll and goodwill, and what they bought
- Group employees
- 4,620
- Asakage Construction, March 2026; about 14,000 partner-firm workers on its sites on a typical day
- Average monthly overtime of site engineers
- 34 hours
- FY2025, from gate and log-on records; 71 hours in FY2018
- Annual consolidated personnel expense
- ¥46.2bn
- FY2025; not the HR budget
- Craftsmen employed directly
- 212
- Asakage Craft Works, March 2026; average age 31
Decisions & consequences
01Pricing working hours into the tender before the limits applied
In 2018 our site engineers worked an average of 71 hours of overtime a month, measured from gate and log-on records, with peaks above 110 hours in the months before a handover. Only 23% of our building sites closed on Saturdays. The statutory overtime limits would apply to construction from April 2024, and our hours could not meet them through HR measures alone. The construction period agreed at tender fixed most of the overtime before anyone was assigned.
In 2019 I proposed that from April 2020 every bid above ¥1 billion be priced on a schedule that closed the site eight days in every four weeks, and that HR sign a page in the tender showing the staffing and hours the schedule implied. Our head of building construction opposed it in writing. He expected competitors to offer shorter periods and clients to find us inflexible. Our chief financial officer worried about margins on fixed-price contracts already signed. The president accepted the proposal for new bids and asked for a quarterly report on tenders affected.
We gave up work. Over the next two years we declined or lost nine tenders, worth about ¥62 billion, where the client would not accept the longer period. Accepted schedules lengthened by an average of 6%. We added a second engineer to sites above ¥5 billion, opened a document support centre in Nagoya with 140 staff to take drawings, photographs and inspection records off site offices, and raised partner-firm labour rates by about 4% so that crews would not lose income when Saturdays closed.
By the year to March 2026, average overtime for site engineers had fallen to 34 hours a month, 81% of building sites closed eight days in four weeks and voluntary turnover among site engineers had fallen from 7.8% to 4.1%. No engineer exceeded the annual limit in FY2024 or FY2025.
I got the sequence wrong. In the first year many partner firms still paid craftsmen by the day, and their workers lost close to a tenth of their income when Saturdays closed, months before our rate increase reached them. Two formwork firms lost young carpenters to competitors. I should have fixed partner pay first. I also failed to count hours in the document support centre for six months, and some of the overtime we had removed from sites had simply moved there.
Taking age out of who leads a site
In 2021 an Asakage site engineer reached the grade that allowed him to lead a site at an average age of 44. Pay rose with age and tenure until fifty-five, and a thirty-two-year-old engineer who had in practice run a ¥3 billion warehouse project was paid on the same scale as a contemporary who had not. Voluntary turnover among engineers aged 25 to 34 had reached 9.2% a year, and exit interviews named developers, consultancies and design firms offering responsibility sooner. Our engineering workforce was also ageing: 38% of site engineers were over fifty.
I proposed replacing our age-and-tenure table with five role grades defined by the scale and complexity of the site an engineer had led or could lead, assessed on evidence: projects run as deputy or manager, the first-class qualification in construction management, safety records and a review by two site managers from another branch. Our employees' union opposed the first draft. Its chair, Tetsuya Morooka, argued that older engineers would see their pay frozen while younger colleagues overtook them, and that assessments would invite favouritism. Our head of civil engineering preferred to raise starting pay and leave the structure alone.
We negotiated through most of 2022. I gave up any saving the reform might have produced: every employee over fifty kept their pay for five years whatever their grade, at a cost of about ¥1.1 billion a year. The union secured an appeal panel with a union member on it and an annual review of assessment outcomes by age band. The new structure took effect in April 2023.
In the three years since, 41 engineers under forty have been appointed to lead sites, against six in the previous five years. Turnover among engineers aged 25 to 34 fell to 5.6%. Average pay for site engineers rose 6.8%, and total personnel expense rose by ¥2.6 billion a year. Appeals ran at 3% of assessments in the first year and 1% in the second.
I would change the grade definitions. They described site leadership well but undervalued estimators, structural checkers and safety officers, whose work does not map onto the size of a site. Eleven of them left in the first year, and we had to add a specialist track in 2024 that should have existed from the start.
Employing craftsmen instead of only buying their work
By 2020 the average age of craftsmen recorded on Asakage sites was 51, and our partner firms in formwork and rebar told us they could not find apprentices. Many relied on technical interns from Vietnam and Indonesia who left at the end of their terms, often just as they became productive. On two hospital projects that year, formwork shortages delayed concrete pours by several weeks. Our order book assumed crews that our partners could no longer promise.
I proposed a subsidiary that would employ craftsmen directly on monthly salaries, with a three-year training plan, Japanese and foreign recruits on one pay scale and a defined route to foreman. Our partner-firm association objected strongly. Its chair, Kazuo Ninomiya, argued that a general contractor employing craftsmen would compete with its own subcontractors for scarce workers and for work. Our head of procurement estimated that directly employed crews would cost about 12% more than subcontracted ones.
We accepted both objections in part. Asakage Craft Works, established in October 2021, agreed not to recruit from partner firms for three years, and its crews work alongside partner crews rather than replacing them. We accepted the higher cost and opened our training yard in Komaki to partner-firm apprentices without charge. Foreign recruits joined on specified skilled worker status, paid the same as Japanese colleagues at the same certified skill level, with Japanese lessons during working hours and housing we leased directly rather than through an intermediary.
By March 2026 Craft Works employed 212 craftsmen with an average age of 31, including 68 from Vietnam, Indonesia and the Philippines. Of the first two intakes, 89% remained after two years. Fourteen have become foremen, four of them foreign workers, and nineteen foreign workers have reached the higher skill level that lets them stay longer and bring their families. Partner-firm apprentices have taken 340 places at the training yard.
What I would do differently is the start of foreign recruitment. For the first intake we relied on a sending organisation whose charges to workers we had not examined closely enough, and several workers arrived in debt. We repaid what we could establish had been charged, about ¥9 million, and now audit recruitment costs in each sending country before we take anyone on.
Letting any engineer choose a home region
In 2017, when I became head of human resources, women were 44 of our 1,380 site engineers, about 3%. Their median tenure was five years, and in the previous three years 19 had left, most between their fifth and eighth year. Exit interviews mentioned site facilities, but the reason given most often was the transfer list: an engineer could be assigned to any project between Shizuoka and Kanazawa at a few weeks' notice, for the life of the project, and the list was drawn up by the construction division to fit projects, not people.
I proposed a home-region declaration open to every site engineer: an engineer could name one of our six regions and be assigned only within it for up to six years, renewable once, with no effect on pay or grade. We would also set a facilities standard for site offices and staff larger sites so that two engineers could share the manager's duties across shifts. The construction division objected that it would be unable to staff remote projects. Several senior women engineers objected too. They did not want a measure that colleagues would read as a concession to women, and I agreed with them, which is why the declaration was open to everyone.
The cost fell on the construction division. Remote projects needed engineers on temporary assignment with allowances, or engineers hired on fixed terms, adding about ¥380 million a year. For two years we bid less often for isolated civil projects in mountain districts.
By April 2026 women made up 151 of our 1,450 site engineers, and 17 women led sites, against three in 2017. Men made 62% of the 410 declarations, most often to care for children or parents. Voluntary turnover among women engineers fell from 11% to 4% a year, close to the rate for men, and the number of engineers leaving to join competitors in their home towns fell by more than half.
What I would change is the pace of the facilities standard. It took three years to reach every site because we negotiated it project by project, and a single date with a budget would have been fairer to the people waiting. I also wish we had recorded exit reasons systematically from the start. We argued about the transfer list for a year on anecdote before we had numbers.