My bank is intermediate small bank with an asset of around 1 billion dollars.

The Consumer Compliance Dept. used to monitor 100% of Home mortgages in regards to Flood, RESPA, TIL 10% tolerance verification and etc.

We are planning to reduce the monitoring scope by only looking at samples.

I wonder if there is any guide on what adequate compliance testing sample size is.

Thank you.

Well, there are two methods. Statistical sampling if the population to large enough. Confidence rates of 90-95% are usually chosen. Rate of occurrence around 5%.

If the population is not large enough to choose a sample statistically, it basically comes down to a judgment call. 10% - 20% is usually acceptable depending on the type of testing you are doing. For example, if the chances for error are limited by software or input errors, 10% is plenty. If there is more subjective processes involved, a higher sample size might be desirable.

There are some pretty decent free resources found here:

http://www.auditnet.org/audit-library/sampling-resources-for-auditors

cgd, I would agree 110% with Randy. One thing that might also be included in the plan is to increase the sample if errors are detected.

Some shops break up their testing into technical and substantive. Use a low number of samples and procedures for technical testing and increase the numbers and details if errors over a certain threshold are found (substantive).

Good points. Historical error ratios and past identified problems are another component of judgmental sampling.

Increasing sample sizes if you find errors however in all cases may not be required dependent on the type of error found. If you identify a systemic error - testing more isn't really going to tell you much more than you already know by testing one or two.

I've seen outside firms pull that one. They set an arbitrary error ratio that triggers an increase in sample size. They have an error ratio of 100% so they then test more. Ka-ching goes the cash register.........

Thank you all for your help.

Just want to mention the reason why we choose to reduce the size of sampling.

We used to review 100% of flood and home mortgages, because we only have about 100 applications a year.

However, reviewing all may hold my department liable for errors, which explains why we would like to select samples rather than reviewing all.

Thank you.