One of the largest challenges facing accountants is working with so much data. It can sometimes be difficult to recognize trends and what is and isn’t important. The sheer volume of transactions challenges the auditor,
especially in cases of fraud where the client has every incentive to compromise records. The profession has had its share of controversy over the past several years not only due to ethics, but because of auditor judgment in the
face of the mountain of data supplied, even from the smallest companies.

How can the profession cope with this? Recent legislation and regulations may help. They do demonstrate to the public that the profession is addressing its issues. However, it is questionable if these reforms meet the real challenge, which is to minimize the technical aspects of an audit and allow more time for the judgment components. If auditors spend too much time on the rote parts of an audit, like calculating expected values, then little time is left for a broader analysis of the business.

Certain aspects of the estimation process can be improved upon through technology. One of these is to set expected values in the analytical procedures portion of an audit or review. “Expected values” are the account values that might be anticipated, based on a company’s past performance. Setting expected values takes time, effort and knowledge of statistics.

Why is this so important? By identifying what general ledger account values should be, auditors can apply a management-by-exception basis for looking into values or accounts that are “out of line” with what they should be. If auditors have to dissect all income statement and balance sheet accounts with equal energy, there is an increased chance of missing larger problems. Setting expected values isn’t perfect, but the methods of deriving values can and should be improved through technology.

Until recently, deriving expected values has been time consuming. For example, say you were attempting to
derive an expected value for sales/revenue for the
following company:

Example 1

What might an “expected value” be for sales of this company in 2005? Here, many methods might be applied to determine expected sales value. However, it looks like the expected value might be $80M. This number is derived by recognizing the sales trend is increasing by 100 percent annually. The data in this example would be a compelling starting point because it is apparent and does not deviate, and there is a clear trend for 2005.

So, in examining data like this, in order to set values, weight the clear financial trend less than the externalities of the firm, market or industry. Take another company example:

Example 2

Again, it may seem logical that the expected value of the company’s sales would increase, but the calculation would be more difficult, given the uneven data for the periods under consideration.

It’s rare when data is as smooth as in Example 1. Where a simple trend analysis might be a good starting point for Example 1, a different mathematical technique (like regression analysis) might be more appropriate
for Example 2.

Regression analysis is cryptic and time consuming, especially when applied to all accounts. While it is possible that you may apply regression analysis to a particular set of company data, think of how time consuming it would be to perform calculations manually against all accounts.

However, technology makes calculating expected values quicker and more reliable in cases where auditors are using both trend (two periods of data) and regression (three or more periods of data) analysis. Mathematical models can be easily applied to these analytical procedures to give accountants time to review overall conditions, rather than devoting time to rote calculations.

If setting expected values is achieved more quickly, how will this specifically help enhance the work done in analytical procedures? Using good statistical modeling does not eliminate the need for high quality data analysis and other methods to determine expected values.

This is where determining expected values becomes more “art” and less “science.” The goal is to minimize the science by automating the calculations and increase the amount of time available for qualitative review/judgment decisions. A reduction in the time it takes to calculate the statistical rote part of the process will allow for more time and research into other components that can make the difference between a good audit/review and a bad one, such as:

• Interviewing management. Understand how financial and economic conditions are changing in the company.
• Performing industry research. Internal factors (management, products, customer service) and external factors (industry changes, employment landscape, government regulations) drive the company’s financial performance.
• Evaluating trends and looking more closely at
  unusual accounts.

Several software applications facilitate the process of setting expected values. Many auditors also use Excel, and macros can be developed that use generally accepted statistical models. Any good system should have a set of algorithms that automatically calculate expected values using historical values.

About the Author
Brian Hamilton is the developer of ProfitCents. Based in Research Triangle Park, NC, ProfitCents produces narrative text on what financial statement numbers and metrics mean in plain language. Hamilton can be reached at brian.hamilton@profitcents.com.

Top