BAD debts are a bank's worst nightmare. Yet they come with the territory. Just like a factory knows that some of its output will come out defective, a bank knows that some of its loans will go bad.

The art of management is not eliminating error but anticipating it, managing it and pricing for it.

So a critical task of the management of a bank is getting the error rates right. And last week Absa showed it had got things rather wrong on one portion of its home loan book. That led to a 9% smack to its share price and much angry criticism of the bank under CEO Maria Ramos.

Should she be blamed?

The "yes" camp has it that she should have spotted the under-provisioning earlier and that she blundered in cutting provisions in last year's results. The rather less populated "no" camp has it that Ramos has shown her willingness to take bad news on the chin when it crops up in her bank and should be applauded for transparency.

The facts are somewhere in between. Ramos has decided to dish out the dirt and swallow the medicine. And yes, she could have been more attentive to the problem earlier, but it is wrong to pin the failure on her shoulders.

It helps to look at the problem in a little more detail.

Banks determine their provisioning levels by consulting a model which tries to forecast what the default rate is going to be. Of course, such models sometimes serve little more than to provide a false sense of certainty.

A provisioning model looks at current repayment rates as well as leading indicators such as gross domestic product growth and employment.

Absa's model has been spitting out the wrong number for its so-called "legal" home loan book. This is the portion for which the bank has entered into legal proceedings to recover its money.

Absa has had a provisioning coverage ratio of 17%. That compares to 28% on Nedbank's equivalent book and 20% for both Standard Bank and FirstRand.

This relative underprovisioning, it turns out, was completely unjustified. Ramos should have asked more probing questions.

How big a mistake has been made? Absa's announcement was only that profits would be up to 10% lower than the same period last year, so it takes guesswork to figure out how big a hole is in the book. The legal book totals R14,4bn, which means it has provisions of R2,5bn for write-offs.

The profit guidance implies that it will earn up to R460m less than last year. But the additional provisions it is raising for its legal book will be far more, as profitability elsewhere has improved.

Were it to match Nedbank's provisioning cover rate, it would need an extra R1,5bn in provisions (ignoring growth in the legal book during the last six months). That would probably be too much.

The banks' legal books largely relate to the mortgage binge they went on in 2006 to 2008, when loans of up to 110% of the value of homes were made.

Nedbank was particularly enthusiastic and as a result had the biggest hangover. Absa was less of a glutton. So at a guess an extra R1bn in costs for extra provisions is coming when it announces results later this month.

This is an expensive stuff-up, and the share price fall on the announcement wiped out R9bn from the value of the bank. That is a vote of no confidence in management (and partly a realisation by the market that bank share prices are too high) rather than a pricing adjustment for the unexpected loss.

There is natural investor apprehension after the spate of departures from the top team at Absa, including last year's departure of retail head Gavin Opperman and deputy CEO Louis van Zeuner, who leaves at the end of this year.

But decisions that affect the book's performance now were made in the pre-Ramos era, when loans were dished out.

The model's underprovisioning apparently only came to light in the last couple of months as write-offs spiked.

Ramos should have spotted the problem earlier, but when it emerged, she chose to take it on the chin.

It's not the first time - shortly after joining the bank she decided to write off R1bn stemming from the single stock futures mess. She then tightened the provisioning on the bank's retail books to recognise loans as nonperforming earlier than before, bringing the bank into line with the rest of the industry. In both cases she took a more conservative stance than many in Absa's senior management believed was justified.

The only certainty is that the market reaction to the announcement last week shows a lack of trust in Ramos. By announcing the bad news quickly, she earns trust to some extent, but a belief that she should have spotted it earlier will count against her. She can ill-afford any more surprises.

With a new senior team at the bank, it is time for Absa to start delivering consistently.