The Bank That Buys Back Its Shares, Then Stops, Then Buys Back More

Here is the basic rhythm of HSBC's shareholder returns over the past year or so: announce a large share buyback, execute a portion of it, take a charge somewhere on the balance sheet, pause the buyback to rebuild a capital buffer, let the buffer refill, then announce another buyback.

It sounds circular because it is. The competitive title you may have seen - "HSBC Announces Fresh $1 Billion Share Buyback After Profit Beats" - is the sort of headline this machine generates every few months. The actual story is less about any single buyback and more about what the whole cycle reveals about how a global systemically important bank manages the gap between regulatory capital and investor expectations.

The capital ratio gate

Banks are not ordinary companies. They cannot buy back shares whenever they want because regulators require them to hold a minimum amount of common equity tier 1 (CET1) capital - essentially the highest-quality, loss-absorbing equity a bank has, measured as a percentage of its risk-weighted assets. If the ratio falls below the regulatory floor plus a conservation buffer, buybacks stop. Full stop.

HSBC's stated medium-term target for its CET1 ratio is 14% to 14.5%. As of the first quarter of 2026, the ratio was 14.0%. That is the bottom of the target range. In March 2026, HSBC told investors it would not resume buybacks until capital ratios improve. In October 2025, the bank paused buybacks for three quarters to rebuild after a series of impairment charges.

The simplest model is this: HSBC generates profit, which flows into CET1 capital. When CET1 sits comfortably above 14%, the bank authorizes buybacks, which reduce CET1. When CET1 approaches the floor, the buybacks slow or pause. Profit accumulates again. Repeat.

It is not exactly a broken system. It is the standard way a large bank returns capital when the regulatory ceiling and the investor expectations floor are close together. But the repeated cycle of announce-pause-resume makes it worth understanding, because the headline framing always suggests the buyback is a fresh sign of confidence rather than the mechanical output of a balance sheet that is generating just enough surplus to fund periodic repurchases.

The numbers behind the headlines

In February 2025, HSBC announced a $2 billion buyback alongside full-year results. In April 2025, it announced another $3 billion buyback after quarterly results that showed lower profit. In July 2025, pre-tax profit plunged 29% year-on-year, to $6.3 billion, on a $2.1 billion writedown of its stake in China's Bank of Communications and rising credit losses in Hong Kong real estate. HSBC announced yet another $3 billion buyback on the same day.

Then, in October 2025, the bank hit the brake: a three-quarter pause to rebuild capital. By the time Q1 2026 results came out in May 2026, the CET1 ratio had settled at 14.0%, the bottom of the target band, and management was clear that the ratio needed to move higher before buybacks could meaningfully resume.

The full-year 2025 picture is instructive. Reported pre-tax profit fell 7.4% to $29.9 billion, dragged down by $4.9 billion in "notable items" - including the Bank of Communications writedown, $1.4 billion in legal provisions, and $1 billion in restructuring costs. Strip those out, and adjusted pre-tax profit rose to $36.6 billion. The adjusted return on tangible equity (RoTE) hit 17.2%, and management raised the forward target to "17% or better" through 2028.

So the profit beat the headline references is real, on an adjusted basis. Adjusted profit grew 7% at constant currency. Revenue rose 4%. The wealth business is the engine. Net interest income increased $2.1 billion on higher yields from reinvested structural hedges and deposit growth. None of this is bad. But the underlying pre-tax profit declined. The charges were real. The capital hit was real.

Why the buyback cycle matters

The interesting question is not whether HSBC is profitable. It is. The interesting question is what form of capital return a CET1-constrained bank should use, and what the repeated buyback-announce-pause cycle signals about the bank's actual capital position.

Dividends and buybacks are not economically equivalent for a bank. Dividends are sticky - investors treat them as semi-permanent, and cutting one sends a negative signal. Buybacks are optional and reversible, which gives management flexibility. But from a CET1 perspective, both reduce common equity by the same amount. A $3 billion buyback and a $3 billion dividend each shrink the CET1 numerator by $3 billion.

The reason banks prefer buybacks is the flexibility: they can slow down or pause without formally cutting a dividend. But the cycle of announcing large programs, executing part of them, then pausing is the visible output of a bank that is generating capital surplus, just not a lot of it. HSBC's CET1 ratio hovers at the bottom of its own target range because the surplus profit each quarter is roughly equal to the amount investors expect to see returned.

There is also a secondary incentive layer. HSBC's stock surged 50% in 2025 and climbed another 10% into 2026. Each new buyback announcement reinforces the price momentum, which in turn makes the stock look like a rewarding investment at a time when many global banks are mired in net interest margin compression. The buyback is both a capital-management tool and a signaling device. Both functions are legitimate, but it helps to see both of them at once.

The Bank of Communications problem

One charge keeps appearing: the investment in Bank of Communications, a Chinese state-owned bank in which HSBC has held a stake since 2003. HSBC took a $3 billion charge on BoCom in early 2025, then another $2.1 billion in the second half of 2025, including $1.1 billion from dilution when the Chinese bank did a private placement. That is $5.1 billion in writedowns over roughly twelve months on a single investment.

The basic point is that this is not a normal portfolio fluctuation. It is the slow-motion recognition that a decades-old strategic investment in a Chinese state bank has a market value far below its book value, and the accounting is finally catching up. Every time HSBC reports results and this investment shows up as a notable item, it is a reminder that a material chunk of the bank's equity was allocated to something that doesn't generate a controllable return, doesn't integrate with the core business, and may never recover its historical cost.

This does not make HSBC a bad bank. But it does mean that the capital surplus driving the buyback cycle is smaller than the adjusted numbers suggest, because the BoCom drag is recurring, not one-off.

The structural judgment

The competitor headline frames any new buyback as proof that the bank is confident and profitable. That is not wrong, exactly. It is incomplete in a way that changes how you should think about the stock.

HSBC is generating capital surplus within a narrow CET1 band. It uses buybacks to return that surplus because they are flexible and because they signal confidence to the market. It pauses them when charges - real charges, on real exposures in Hong Kong property and Chinese banking - eat into the buffer. Then it starts again.

The machine works. The question for an investor is whether the surplus being generated is large enough and durable enough to support the return profile management is promising - 17% RoTE or better through 2028, rising revenue, and a cost-cutting program that is supposed to save $1.5 billion annually by end-2026 - while still keeping CET1 above 14% when the next unexpected charge arrives. It is a solvable problem, but it is a narrow corridor, and the repeated pause-and-resume cycle is the visible proof of that.

If you own the stock, the buyback is a floor, not a thesis. The thesis is whether the adjusted profit growth in wealth and net interest income is durable enough to widen the CET1 band over time, or whether the bank will continue to dance at the bottom of its own target range, buying back shares when it can and pausing when it has to.