If auditors won’t audit, mandatory rotation won’t help. Part 3

I have a discussion here and here about whether the halo effect on auditors is so strong that mandatory auditor rotation is needed to get good audits.

How the design of audit helps counter the halo effect

After reading much of Mr. Kahneman’s book, I would suggest that an audit is designed to counter the System 1 / System 2 biases.

Consider the initial analytical review. The goal is to take a first look at the financial information to see if there is something different from last year that needs extra attention or if there is anything obviously wrong in the numbers.

What could be odd? Maybe the gross margin or GM percentage is too good.  Maybe inventory turnover is way up in a lousy economy.  Payables may be low when everyone else in the industry is stretching a/p.  Maybe the allowance for uncollectible receivables or loan losses is the exact same as last year.

If something is odd, the auditor knows to focus on that relationship.

On the other hand, maybe the numbers all make sense and are sort of in the range of what you would expect.

Everything’s fine you think, right? That’s system 1 at work – thinking fast.

So, is the audit done if the initial analytical review makes sense when thinking fast?

No way.

There is an extensive risk assessment that needs to be done. Discuss and document entity level controls and activity level controls.  Ponder and discuss where fraud could be hiding.  Evaluate design of internal controls, test whether they were placed in service and then test the controls that will be relied upon. Then there are lots of further audit procedures.  Confirmations, vouching, more interviews, tieing out, predictive tests.  Evaluate each substantive test to make sure it was what it should have been.  Then wrap-up interviews (contingencies, subsequent events, etc), final analytical review, painfully long disclosure checklists, and final internal review. Cold review for SEC work.

All of those steps fit into the System 2 of Mr. Kahneman’s framework. Think slow.

Each of those steps ought to give an auditor opportunity to challenge whether the accounting and explanations and numbers make sense or whether there is something that just doesn’t seem right.  In his model, that ought to provide time to ‘think slow’.

If you try, those procedures ought to overcome the halo effect and shake out material misstatements in the financial statements.  System 2 is built in to the process. 

On the other hand, if you’re not trying, all that think slow stuff won’t work.  If you are still thinking fast while performing additional procedures it won’t work.  If the next firm isn’t trying either, mandatory rotation won’t do any good.

Lesson learned for me about halo effect and decorrelation of error

I even see an effort to decorrelate errors in the professional literature. When performing a predictive test, either in an audit or a review, the accountant is supposed to establish an expectation for the test and then prepare the predictive test.

I think we CPAs oftentimes do that in the wrong order in light of the authoritative literature – we do the test, then figure out our expectation, then reconcile the two.

I found myself doing that recently on an engagement for a church.  Here’s the situation:

Contributions are extremely close to flat, up just a bit after dropping the previous two years. Okay, thinks me, that makes sense in this economy. Oh, yeah. Gotta’ check what happened to attendance – oops. Get the data.  Okay, attendance is flat, within the accuracy of the client’s statistical tool. Okay, flat attendance means giving should be up a bit, so the contributions are reasonable.

That’s the wrong approach, in light of authoritative literature.

(In my own defense I compared that giving-compared-to-attendence relationship to other information about the client and to my experience from my other clients.  The relationship still made sense.)

That is also the wrong order in light of the halo effect. We should decorrelate the error. Figure out what the change or ratio should be before we even look at the current year number.  That way the actual results don’t taint our conclusion about the actual results.

Lesson learned for me – make extra sure I prepare my expectation before performing a predictive test on any audit or review.

Next post is on counter pressures to doing a lousy job on the audit and how the counter pressures sometimes don’t work.