59% of Foreign Subsidiaries in Japan Changed Heads, 2002-07

Head offices treat replacing the Japan head as a fix. The numbers haven't come, so the person must be wrong, and a better person will turn it around. The one large study I could find on this says otherwise. It covers 554 long-established foreign subsidiaries in Japan between 2002 and 2007, so it describes that period and that kind of company. In that sample, when an expatriate head was replaced, the subsidiary's profitability went down afterward. It didn't matter whether the successor was Japanese or another expatriate.

Who is in this number?

Established foreign companies, and no startups. The researchers are Fabian Froese, then at Korea University, and Ralf Bebenroth of Kobe University. Their data come from Toyo Keizai's directory of foreign-affiliated companies in Japan. The 2012 discussion paper covers 554 subsidiaries and 2,113 firm-year observations, with 521 changes of the top manager between 2002 and 2007.

The paper describes the sample plainly. The average subsidiary had 835 employees, was 29 years old and was about 77% foreign-owned. The analysis needs performance data for three years before and after each succession, which is why the successions are limited to 2002 through 2007.

Two cautions follow. A Japan entity that is two or three years old with a dozen people is not in this data, and I haven't found public statistics that count such entities separately. The data are also twenty years old. Read them for the pattern in that period and that kind of company. Don't read 59% as today's rate.

Do Japan heads really get replaced in year two?

I couldn't find data that says so. Between 2002 and 2007, 59% of the subsidiaries changed their head at least once and 25% changed more than once. Turn that around: 226 of the 554, about four in ten, recorded no change at all in that period. The paper does not report an average tenure. It only notes that an expatriate's term is usually three to five years.

For comparison, the search firm Russell Reynolds Associates tracks CEO turnover at companies in the major stock indices worldwide. CEOs who left in 2025 had served 7.1 years on average, down from 8.3 years in 2021. The two figures count different things and can't be compared directly. As a rough scale, an expatriate's three to five years is much shorter than a listed-company CEO's tenure, and it still isn't two.

So where does the year-two impression come from? My guess is two clocks that don't line up.

The first clock belongs to the person. In The First 90 Days, Michael Watkins asked more than 200 company CEOs and presidents how long a typical mid-level leader in a new role takes to reach the break-even point, where the value contributed equals the value consumed. Their average estimate was 6.2 months. That is for an ordinary transition, with no change of country or language. Froese and Bebenroth, in an illustrative example of a three-year assignment, assume the first 12 to 18 months go to adjustment. That is an assumption for the example, not a measurement.

The second clock belongs to head office, and it runs on fiscal years. The first full year of Japan numbers lands on the table roughly a year to a year and a half after the person arrived.

Put the clocks together. The first serious review arrives just as the head has finished adjusting and started to produce. I suspect many of the 25% that changed heads more than once in six years stumbled at this point. That is inference. No statistic says it.

Half of all successions were Japanese to Japanese; profitability fell after the two kinds where an expatriate stepped down

521 top manager successions at 554 foreign subsidiaries in Japan, 2002 to 2007, by who replaced whom (%). Froese and Bebenroth (2012, Kobe University RIEB discussion paper), data from Toyo Keizai's directory of foreign-affiliated companies

• Japanese to Japanese261 cases
  50.1%
• Expatriate to expatriate142 cases
  27.3%
• Japanese to expatriate67 cases
  12.9%
• Expatriate to Japanese51 cases, localization
  9.8%

Predecessor was an expatriate: the paper reports lower return on sales after the change

About four in ten subsidiaries kept the same head throughout. After an expatriate stepped down, margins fell.

Half the handovers were Japanese to Japanese

Of the 521 successions, 261 were Japanese to Japanese and 142 were expatriate to expatriate. In 67 an expatriate replaced a Japanese head. In 51 a Japanese manager replaced an expatriate.

That last kind is what people call localization. It was the rarest, under a tenth. The reverse move, sending someone from headquarters to replace a Japanese head, happened more often.

"Send our own person to open the market, then hand over to a local" is a plan you hear a lot. Among these long-established foreign companies, few had carried it through.

After an expatriate left, margins fell

In this sample, when an expatriate head was replaced, subsidiary performance declined, whoever the successor was. That is the conclusion of the version the two authors published in the Journal of International Management in 2020. In the 2012 discussion paper, performance is measured as return on sales.

The abstract adds two things. If the subsidiary was already doing badly, a newly sent expatriate may not be able to stop the decline. And replacing an expatriate with a local manager can improve performance, mainly in older subsidiaries.

In the 2012 discussion paper the authors suggest that expatriates on short terms may be short-term oriented, for example holding back investment in R&D and training. My own reading is plainer. The change itself costs something. On the page about building a Japan entity from zero I wrote that replacing a wrongly hired Japan head takes six to twelve months, and the Japan business more or less stops in between. Whatever the last person took a year or more to learn, the next one learns again.

One caution. Companies that are doing badly are the ones most likely to change their head, so poor results afterward can't all be blamed on the change. The paper controls for prior performance, but it is observational data. It shows association.

What can head office write down before the appointment?

What to look at and when, what Japan can decide on its own, and what the role actually requires. If these are written before the appointment, the year-two review has something to be checked against.

Timing first. What should be visible at 6, 12 and 18 months? For the first year I argue for milestones: how many channel partners are signed, whether there is a first customer case that can be made public, whether the service commitment has been met for several months running. Measure someone who has only just passed break-even by first-year revenue and you will mostly measure disappointment.

Then authority. Up to what amount can Japan approve by itself, and which terms can it negotiate? Put it in numbers. Without that, a review can't tell "didn't do it" from "wasn't allowed to." Who decides when the boss is away belongs on the same page.

Then the role. I think many successions are set up on the day of the appointment. Promoting your best salesperson to country manager is the standard example. If what the role needs and what the person is good at don't match, two more years won't change the result.

A change at the top also leaves one thing untouched. The habits underneath stay. What the first five people settled lives on in everyone who remains, and the new head inherits the same organization.

The study is old and it can't tell you whether your Japan head is the right person. What it found, in those companies and those years, is that replacing an expatriate head was not free. The bill arrived as lower margins.

---