Set the payout from the stress test, not the share register
When a bank sits inside a listed holding company, its dividend is often settled by what the parent has already told its own shareholders. The payout should instead be whatever is left after a capital plan has survived a severe scenario.
There is a common structure in Malaysian banking, and in much of Southeast Asia, in which a licensed bank is wholly owned by a listed holding company. The holding company's shareholders want a dividend. The holding company has little income of its own except the dividend it receives from the bank. So the question of how much the bank pays its parent is, in practice, the question of how much the parent can pay the market. Most of the time the arrangement works without strain. When it strains, the bank's capital is what gives way.
My argument is simple. The bank's payout should be set by its capital plan after a severe stress scenario has been run through it, and the parent's dividend policy should follow from that figure. The order matters. A payout settled first and tested afterwards will tend to pass the test, because the people running the test know which answer is wanted.
A promise made before the capital plan
In 2021 our parent, Sejahra Holdings Berhad, told its shareholders that it aimed to distribute about 60 per cent of earnings. The aspiration was reasonable on the figures available that year. Financing growth was modest, credit costs were falling from their pandemic peak and our capital ratio sat comfortably above the board's internal floor. Nobody, including me, asked the bank to model whether 60 per cent would still be affordable if growth picked up and a downturn followed.
By late 2023 the question could no longer be avoided. Our plan assumed financing growth of 8 per cent a year, and the risk function's severe scenario combined a sharp rise in SME defaults with a fall in property values and a period of deposit outflows. Under that scenario, a 60 per cent payout took our CET1 ratio to 11.4 per cent at its trough, below the board's internal floor of 12 per cent. That floor is well above the regulatory minimum, including the capital conservation buffer, and breaching it would not have been illegal. It would have meant that, in a bad year, we would be cutting financing to rebuild capital exactly when customers most needed us to keep lending.
I recommended 40 per cent. The parent's managing director argued for 60, and his argument deserved a hearing. Markets reward consistency in dividend policy; a holding company that cuts its payout without warning is assumed to know something bad. Two directors appointed by our largest shareholder supported him. The board settled on 45 per cent for 2023, combined with a reinvestment plan. The share price fell 9 per cent in the following week.
Reinvestment is not a disguise
A dividend reinvestment plan lets shareholders take new shares instead of cash. In our case, the parent offered its shareholders that choice, and the cash it kept because 61 per cent of its shares elected for new shares was subscribed back into the bank as new ordinary equity, RM107 million in all. Some investors described this to me as a capital raise in disguise. I think that criticism is half right and worth taking seriously.
It is right that reinvestment retains capital that a cash dividend would have released. It is wrong to call it a disguise if the bank says plainly why it is doing so. Our announcement stated the stress-test result, the internal floor and the amount the plan was expected to retain. Investors who wanted cash could take cash. Those who elected for shares were choosing to fund the bank's growth at a known price. Where reinvestment becomes a disguise is when it is used to keep a headline payout ratio unchanged while the bank quietly depends on take-up it has not disclosed. A bank that needs reinvestment to meet its capital plan should say so, and should show what it would do if take-up were low.
The results so far support the approach, though two years is not a cycle. CET1 rose to 13.1 per cent at the end of 2024 and 13.6 per cent at the end of 2025. The stress-test trough is now 12.6 per cent. We have issued no additional tier 1 capital, which saves a coupon we would otherwise be paying for years. The payout for 2025 was 50 per cent, and the parent has told its shareholders that future payouts will follow the bank's capital plan rather than a fixed ratio.
What a dividend paper should contain
I now write the bank's dividend recommendation as the last page of the capital plan rather than as a separate paper. Directors approve the capital plan first, at a separate meeting, so that they are not deciding the dividend and the plan in the same hour with the same pressures.
The plan sets out financing growth by segment, the capital each segment consumes, the stress scenarios and who designed them, the internal floor and the reason it sits where it does. It shows CET1 and total capital under the base case and the severe case for three years. Only then does it show the payout, as the amount the plan can release while keeping the severe-case trough above the floor. If the parent wants more, the paper states what would have to change, such as slower growth, a capital instrument or a lower floor, and what each of those would cost.
There is an argument that this approach is too conservative for a bank our size. I take that argument seriously, and the floor is reviewed every year. A floor can be lowered when the bank's risks change, or when the board is satisfied that management actions in a stress would be credible. What I resist is lowering it because a dividend has already been announced.
I would offer the same advice to any independent director of a holding company whose main asset is a bank. Ask whether the bank's capital plan was approved before the dividend was proposed, who designed the severe scenario, and what the payout would have been if the plan had been written first. If those answers come easily, the payout is probably sound. If they arrive slowly, the share register is deciding the dividend.