Three in Four IT Engineers in Japan Work for a Vendor

If you want to build an in-house engineering team in Japan, the first number to know is 73.6%. That is the share of Japan's information processing and communications workers who are employed by IT companies: software houses, system integrators, outsourcing firms. Only 26.4% work inside the companies that use the software, in manufacturing, finance, retail and so on. In the United States the split runs the other way, 35.1% to 64.9%. The people you are trying to hire mostly sit on the vendor side of the market.

Who is counted in the 73.6%?

The denominator is everyone in Japan's 2020 national census whose occupation was "information processing and communications engineer," 1,253,930 people. The category covers system consultants and designers, software developers, and other IT and telecom engineers. It counts what a person does now, so a self-taught developer is in and a computer science graduate who manages budgets is probably classified as a manager and left out.

The numerator is the people in that group whose employer is an IT company. The Information-technology Promotion Agency (IPA), the government body that did the analysis, defines IT companies as three industries: software, information processing and provision services, and internet-related services. Everyone else counts as "non-IT companies," which I'll call user companies.

The US figure comes from the Bureau of Labor Statistics' Occupational Employment and Wage Statistics for 2021, covering 4,981,090 people. The IT industries there are software publishers, data processing and hosting, and computer systems design. The years differ by one, one source is a census and the other an employer survey, and the industry lists only roughly match. Treat this as a comparison of magnitudes, not of decimals.

IPA first ran this comparison in its 2017 IT human resources white paper, using 2015 data: Japan 72.0%, US 34.6%. It reran it in the DX White Paper 2023 with the new census and got 73.6% and 35.1%. Five or six years of data, and neither country moved.

Three in four Japanese IT engineers work for IT companies. Two in three American ones work for user companies

Workers in the occupation "information processing and communications engineer," split by employer industry, %. Japan: 2020 national census (1.25 million). US: BLS OEWS 2021 (4.98 million). Compiled by IPA, DX White Paper 2023

User companies (all non-IT industries)IT companies

Japan　2020

• IT companiessoftware, information processing services, internet services
  73.6%
• User companiesevery other industry
  26.4%

United States　2021

• IT companiessoftware publishers, data processing and hosting, computer systems design
  35.1%
• User companiesevery other industry
  64.9%

Your candidate pool for an in-house team is the 26.4%. The rest you will have to recruit out of vendors.

Why is Japan shaped this way?

Because Japanese user companies decided decades ago that IT was something you buy. A bank, a carmaker or a retailer that needed a system contracted a system integrator (SIer in Japanese shorthand), which passed much of the work down to second and third tier subcontractors. The work lived at the vendors, so the people grew up at the vendors. Repeat for forty years and you get 73.6%.

A typical corporate IT department in Japan was designed to manage budgets, manage vendors and write requirements. There was never a seat in it for someone who writes code. The grade table, the pay table and the promotion path were all built for a procurement function.

Are they short of people? Very. In 2022 IPA surveyed 543 Japanese and 386 American companies. Asked whether they had enough people to drive digital initiatives, 10.9% of Japanese firms said yes, against 73.4% of American firms. 49.6% of Japanese firms said they were "severely short," against 3.3% in the US. They are short and they keep outsourcing, because the internal structure can't hold a senior engineer. The famous "790,000 IT worker shortage" has the same root; I traced where that number came from earlier.

So contract development and staff-augmentation firms (SES) are the skeleton of Japan's IT industry, not a passing phase. When you hire in Japan, you hire inside that skeleton.

Where the candidates are, and how to get them

Start with the order of magnitude. 26.4% of 1.25 million is about 330,000 people in user companies; 73.6% is about 920,000 at vendors. Both are my arithmetic, IPA gives only the shares. The 330,000 includes people whose job is budget control, people who only run operations, and people who have spent twenty years in a large company's IT subsidiary. Nobody has counted how many have actually built software inside an operating business. It is far fewer than 330,000.

So most of your hires will come from the 920,000, from integrators and SES firms. Before you start, a few things are worth settling.

What they can do versus what you want. Vendor-side engineers in Japan are usually strong at delivery: build to spec, ship on the date, leave documentation. Their weak side is the other end. Technology choices were made by the client or a senior, requirements arrived from the customer, and they rarely owned a decision that went wrong. In-house work is exactly that end. In interviews, ask less about what they built and more about what they personally decided at the last job, and what happened when it didn't work.

Your grade table. An integrator's grades are built on billing rate per person-month and headcount managed. After a certain age the only way up is project manager. People who want to keep coding have nowhere to go. If you import head office's engineering ladder unchanged, Japanese candidates can't read it. If you copy the integrator's, you drive away the people you most want. Draw a technical track parallel to the management track, name the grades, and write down what each grade is allowed to decide. It's the cheapest fix with the biggest effect. I wrote about how the shape of the team ends up shaping the product.

Real authority over technical decisions. The most common first-year disappointment for someone who moved from a vendor to a user company is discovering that decisions still sit with head office, or with a manager who doesn't code. They changed employer and kept the same role. That person leaves within a year, and tells their network before they go.

Pricing, and the other option

Pulling people from vendors into user companies usually means paying a premium in Japan. I won't put a number on it, since it varies too much by sector. One rule of order matters more than the amount: the premium goes on top of a grade table, so build the table first. Do it the other way round and your first three salaries will lock the structure for the next thirty.

There is also the route most Japanese companies actually take. Keep the in-house team small, hold a core of a few people, and keep using vendors for the rest. There's nothing wrong with it. Fighting a 73.6% structure head-on is expensive. Working with it, while keeping the key decisions inside, is often the better trade.

What I can't tell you is whether the structure loosens over the next five or ten years. Between IPA's two measurements, Japan's share moved from 72.0% to 73.6%, in the wrong direction. AI lets small teams do more, which should favor keeping people in-house, but no one has measured how heavy Japanese corporate procurement habits are. The same habit I describe in Japan by the Numbers applies here. When someone tells you engineers in Japan are easy or impossible to hire, ask whether they mean the 920,000 or the 330,000.

---