Six years is long enough to run a provincial branch
Long-serving provincial general managers bring knowledge and relationships an insurer values. They also accumulate arrangements nobody else understands. A fixed tenure limit, with honest support for the families who move, protects both the branch and the leader.
I think a life insurer in China should limit the time any general manager runs the same provincial branch, and that six years is about right. I hold this view while admitting that some of the best branch leaders I have worked with stayed much longer, and that their provinces did well under them. My argument is not about any individual's integrity. It is about what happens to a branch when its working arrangements come to depend on one person's relationships, and about what an insurer owes the people it asks to move.
How a branch becomes its manager's
A provincial branch of a Chinese life insurer is a substantial business. At Songbai the largest have more than 1,500 employees, several thousand agents and partnerships with dozens of bank branches. The general manager deals with the local office of China's financial regulator, the provincial leadership of partner banks, the agency's team directors and, often, local government. Much of that work runs on trust built over years, and a capable general manager accumulates a great deal of it.
The difficulty is that trust and arrangements look alike from a distance. In 2023 eleven of our 27 general managers had led the same branch for more than eight years. Our internal audit reports showed that branches led by long-serving general managers averaged 7.6 expense findings per audit, against 2.9 elsewhere. The findings were not about personal enrichment. They were about arrangements: agent training costs that in fact paid bank-channel fees above the rates filed with our products, service agreements renewed without tender, and staff appointments nobody at head office could explain. Each arrangement had made sense to someone at some point. Together they meant that the branch's real economics were known to one person.
China's financial regulator has made clear that the expenses insurers actually pay should match the expense assumptions filed for their products, beginning with bancassurance and extending to other channels. A branch whose fee arrangements rest on one leader's relationships is hard to bring into line with that expectation, because nobody else knows which arrangements exist. A new leader asks why things are done as they are, and the answers are often revealing.
I proposed a maximum of six years as general manager in one province, phased over three years so that no branch lost its leader without a prepared successor. Deputy general managers now gain experience in a second province before promotion. Our president was cautious and asked for evidence that rotation, rather than the extra audit attention that came with it, would change the findings. By mid-2026 ten general managers had moved. Expense findings in their former branches fell to 2.2 per audit, against 3.1 in branches that received extra audit attention but kept their leader. That is a modest comparison drawn from a small number of branches, and I present it as such.
Rotation is a family decision before it is a staffing one
The strongest objection to rotation is human. A general manager in their late forties often has a spouse with a career, children in school and parents who need care. In China, school places, household registration and the university entrance examination make moving a teenager between provinces a serious matter. An insurer that rotates leaders without addressing this is choosing, in effect, leaders without family obligations, or leaders who will live apart from their families.
Our package covers housing in the new province, two return journeys a month, help with school places and support for a spouse's job search. It costs about RMB 19 million a year. It was not enough on its own. In the first round we ignored the school calendar, and two strong general managers with children approaching the entrance examination declined to move. One of them left the company. A one-year deferral for families in that position would have cost almost nothing, and we have since adopted it. I learnt a version of this lesson in Sichuan in 2008, when we sent colleagues into the earthquake-affected counties for months, and I am embarrassed to have needed it again.
Four general managers chose to retire or leave rather than move. Some of them would have served well for several more years. I do not think that makes the policy wrong, but a board should hear it plainly: a tenure limit will cost an insurer some good people, and the company should decide in advance that it is willing to pay that price.
The bank relationship belongs to the company
Our head of bancassurance opposed the policy because he expected partner banks to use each change of general manager to renegotiate fees or move volume to competitors. He was partly right. We prepared by introducing each incoming general manager to the partner banks' provincial heads alongside the outgoing one, and by having head office rather than the branch sign renewed cooperation agreements. That helped, but it did not stop some banks testing the new leader. In the first two quarters after each rotation, bancassurance premium in the rotated branches fell 7%, against 1% elsewhere. Premium recovered within three quarters in eight of the ten branches. In the other two we lost business that we had probably been keeping on terms we should never have offered.
I draw a broader conclusion from that. If an insurer's relationship with a bank survives only while one general manager stays, it is a personal relationship that the company is paying for. Rotation tests which relationships belong to the company. Those that survive turn out to be worth more than they appeared; those that do not were less secure than the premium figures suggested.
Six years is a judgement, not a rule derived from data. It is long enough for a leader to complete one significant change and see its results, and short enough that arrangements do not harden into the branch's operating model. Another insurer might choose five years or eight. What I would urge any board to avoid is having no limit at all, so that tenure ends only through retirement, scandal or a dispute with head office. Each of those is a worse way for a branch to change hands than a date agreed six years in advance.