FOR all its many flaws, Citigroup is never dull for long. At the IMF meetings in Tokyo at the weekend, Vikram Pandit, its chief executive, looked happy as Larry, gladhanding guests at a reception. At the start of this week, the bank announced third-quarter earnings that, given spectacular past losses and modest expectations, might be called pleasantly uninteresting. Only a day later Mr Pandit abruptly resigned; John Havens, Citi’s chief operating officer, left, too. A replacement was hastily found from within the bank: Michael Corbat, previously head of Citi’s businesses in Europe, Middle East and Africa, is a veteran whose tenure dates back to 1998.

Whatever Mr Corbat’s merits, the old boss is still the one in the limelight. In response to a deluge of queries about why Mr Pandit had departed, an analysts’ conference call was hastily convened on October 16th with Mr Corbat and Michael O’Neill, the bank’s chairman. In comments that provided little illumination, they said all was well and all was subject to change. Mr Pandit was granted a shower of faint praise; no explanation was given for why a five-year stint ended so abruptly.

An unexpected end fits neatly with the oddness of Mr Pandit’s Citi career. Brought into the bank as a result of a wildly overpriced acquisition of a hedge fund he ran (which Citi subsequently closed), Mr Pandit had been around for only five months when, on December 11th 2007, he was made chief executive. At the time of his appointment, the conventional view of Mr Pandit was that taken by Sheila Bair, then head of the Federal Deposit Insurance Corporation. “I doubted that he was up to the job,” is how she puts it in her new book. The verdict now that his tenure has come to an end should be more nuanced.

Since his elevation, Citi’s share price has lost 89% of its value, putting him among the least successful of the current crop of bank bosses (see chart 1). Citi’s recovery is far from complete and the bank’s long-term earnings power is unclear. Mike Mayo, an analyst at CLSA, a broker, faults him for misjudging the severity of Citi’s problems and not pruning the scope of its businesses sufficiently hard.

Mr Pandit did not manage to rid Citi of its historic propensity to slip up, either. A request in March to provide a multi-billion-dollar dividend payment to shareholders was embarrassingly rejected by Citi’s regulators as imprudent. The third-quarter results included a $4.7 billion hit on the valuation of its stake in Smith Barney, a retail broker, which the bank is accused of selling too cheaply to Morgan Stanley.

The case for the defence is that he took control of a tottering institution at a time of crisis, and with lots of government help, went some way to making it better. The number of employees dropped from 374,000 to 262,000. Some businesses were divested; bad assets were put in an entity, called Citi Holdings and originally run by Mr Corbat, that is being wound down. From a dreadfully low base, returns and profits both improved during Mr Pandit’s reign (see chart 2).

Whatever the judgment of history, investors were increasingly infuriated by Mr Pandit’s pay. After giving him a token sum for two years, Citi more recently showed largesse that would not have been out of place in a truly successful organisation. His role, in essence, was to run a very messy public utility. His compensation was appropriate for someone taking great personal and professional risks. A $15m payment for 2011 was rejected by shareholders in an advisory vote earlier this year.

What really brought Mr Pandit’s tenure to an end will doubtless emerge in the days to come. The most popular theory is a fundamental disagreement with the board—whether on something as prosaic as Mr Pandit’s compensation or as profound as what the future Citi should look like. Mr Pandit is believed to be an advocate of the bank’s current (still vast) scale; others are said to want a dramatic contraction in Citi’s balance-sheet. One fact, at least, was clear: when Mr Pandit offered to leave, there was no time for goodbye. The separation was instant and complete.

Citi’s shares rose slightly on news of Mr Pandit’s resignation, neither an endorsement of Mr Pandit nor a sign of rapture with his successor. Mr Corbat has a big job ahead of him. He needs to appoint an executive team. He needs to make difficult decisions about asset sales while preserving the institution’s morale. Most of all, he needs a strategy that takes full advantage of Citi’s remarkable global network and sorts out its oddly stunted operations in the United States.

But the pressure on Citi’s chairman is even more intense. Mr O’Neill, who was an industry star in the 1990s at Bank of America and Barclays before stepping away because of health problems, was unable to offer a clear explanation for Mr Pandit’s exit. If Mr Corbat staggers, the question will not be about how to end the new chief executive’s tenure, but rather, after so many surprises and suspect decisions, that of the board.