Does anyone perform direct verification any more? Sending a notice to the customer with their loan or deposit balance and asking them to confirm the balance is correct? If you do this, what kind of cut off do you use? I am taking a random sample of 5% of our accounts, but is that too high?

Some feedback please!

Our external auditors still utilize this audit technique. I have no idea how they pull their sample, but from past experience, I think it's kinda an exercise in futility. The small percentage of people who actually read the things usually call us or bring them in so we can tell them if it's right!

Maybe someone else has a great idea for an audit technique that will accomplish the same purpose, but get a better bang for your audit buck.

Even if you don't get a response the test has some value - if you ask the borrow to confirm their balance at a figure significantly greater than the figure they expect you can be reasonably sure that they will object. If you get no replies disputing the banks' figures you are able to take some assurance that the loan book value is not overstated.

Of course a reply saying "Loan! What loan?" is also informative!

We do confirmations in conjunction with our external auditor. Many times we actually confirm more than they ask for.

Without pulling out the workpapers, I would guess that we confirm the following:

1-2% of the number of deposit accounts; positive on large accounts and public funds, negative on the rest. We will probably confirm a much deeper penetration of dormant accounts in the future, on a positive basis.

25-30% of the dollar amount of loans. Large loans over a certain threshold are positively confirmed and returned directly to the external auditor. Smaller loans are sampled and negatively confirmed. We stratify everything below the large loan threshold into different ranges and take different samples in each (i.e. 10% of the dollar volume for loans between $500M and $750M, 5% of loans between $100M and $500M, etc.). All participations purchased and sold are confirmed.

While all of confirmations provide some benefit / assurance, I believe, from a risk and materiality standpoint, the large loan and deposit accounts, public accounts, participations and dormant accounts are the most important. On your samples, be sure and use statistically valid sampling techniques so you can extrapolate your findings to the entire portfolio.

Our external auditors do negative confirmation of every "n"th loan and deposit account below $125,000, and a positive confirmation of all accounts over $125,000. A % of each portofolio is targeted which drives the "n" factor. However, it is usually around every 5th - 10th account.

We have previously discussed using an aggregate account balance vs. targeting only single account balances. This was dicussed due to the question of effectiveness.

Hopefully controls have been implemented for approval of large lines such as presentation to loan committee, etc., so the likelihood of a single fraudulent large loan is lower. Mulitiple smaller balance fraudulent loans may be easier to "get through" so they should stay on the radar of audit. This next year we are planning on lowering the threshold and using aggregate debt.

I do Positive confirmations for accounts with a higher asset value, such as trust asset accounts. I do negative confirmations on transaction accounts. The sample size is simply based from the number of accounts in our test group. If I select 30 accounts for review, 30 get notices.