Via @truenfairview, I'm led to an article in Economia on the impact of auditor rotation on auditor independence:
Long-lasting relationships between auditors and their clients do not lead to a loss of independence on the part of the auditor, with independent research suggesting the opposite
Independent researchers at the Nyenrode Business Universiteit in the Netherlands have discovered that auditors are more likely to ensure that corrections are made to financial statements if they had had a relationship with the client for at least five years.
However, their research also revealed that the impact of the audit decreases where the firm provides more than 30% of non-audit services to the audit client – although the “important” audit differences were always corrected.
In a survey of information drawn from the audit files of 147 Dutch listed companies, pension funds and very large privately held companies, the researchers found that in 23% of the organisations, the audit led to corrections of amounts involved in the balance sheet and profit and loss accounts before the audit opinion was issued.
The audit also impacted positively on the notes to the financial statements and the directors’ report where corrections were made in 77% of cases.
Professor Leen Paape said that 72% of the important audit differences were corrected in the annual report, while the remaining 28%, concerning 14 organisations, were below the pre-determined error margins.
According to his co-researcher, Joost van Buuren, more audit differences were being corrected where the client relationship was longer than 10 years than when it was shorter.
However, the positive effect on the processing of auditing differences was already there if the auditor had been auditing for a client for longer than five years.
Instinctively, I want to agree with their conclusion. It seems that the risk of being over-familiar with an audit client is just one factor that influences overall audit quality. But I find the researchers' methodology to be very suspect. It's not clear to me that "adjustments initiated by auditors" is necessarily a good proxy for audit quality. For example, it might be that diligent managers are more likely to appoint new auditors. That could mean there are fewer errors simply because the accounts are right in the first place.
Also, the concept of "auditor rotation" is often misunderstood. The term 'auditor rotation' is used to refer to the audit firm changing. But the IFAC Code of Ethics for Professional Accountants states that over-familiarity of key personnel with a client can be a problem, and that the solution will ultimately be for the audit staff affected to rotate off that client. So members of staff already rotate. Similarly, key members of client management may also change on a fairly frequent basis. The idea that, because the audit firm has been associated with the audit client for many decades, there is necessarily a problem of independence is far from obvious when the audit staff and client management are changing frequently. Again, it's unclear from the Economia article whether the researchers' methodology takes account of audit staff and management newness as a variable in their regression analysis. For example, it could be that a change in client management is associated with errors by the new managers, due to their unfamiliarity with their new employer's business and systems.
This is a dangerous area for the big audit firms. Barnier seems hell-bent on demanding fixed rotation periods of six years (or nine years where a joint audit is undertaken). It's still uncertain whether the UK's proposed ten year audit tendering on a "comply or explain" basis for FTSE 350 companiess will be seen as a sensible compromise. The firms need good, solid evidence that they can rely on to support their case that rotation will be bad for audit quality. I'm just not sure that this piece of research will help them much.