U.S. Auditors Shrug Off a Trio of Impending Rule Changes

U.S. auditors have been looking over the shoulders of their financial institution clients for more than a year as banks march toward a transformational change in how they book loan loss reserves.

At the same time, they’re adjusting to a set of new standards for auditing—rules that will shape how they approach estimates like credit losses.

The collision of rules, however, isn’t fazing many auditors, who appear to be taking the changes in stride.

“The challenge is not so much that there is something completely unprecedented happening or that it’s something we’ve never audited before. We’ve audited complex estimates,” said Graham Dyer, a partner in Grant Thornton LLP’s accounting principles group. “Auditing complex estimates are just inherently difficult. We’ll certainly devote significant resources to it.”

The firm is keeping tabs on the progress of its largest clients as they work through the accounting and test their economic models. Staff are digging into the requirements of the new accounting and the firm has added to its in-house model evaluation team — all to ensure they are prepared, Dyer said.

Reviewing loss estimates and any related method changes are a routine part of any audit. Plus, the basic approaches to auditing estimates remain intact despite the rule changes, Dyer said.

The banking industry has been less sedate. Many have fretted over how much extra scrutiny the already complex accounting will face when banks begin reporting under the current expected credit loss estimate for the first time in 2020.

The new audit standards, ironically, aim to reduce subjectivity in financial statements at the very same time a new accounting standard that increases subjectivity goes live, said Michael Gullette, senior vice president at the American Bankers Association.

Better known as CECL, the figure is intended to gauge loses over the life of a loan and can climb to several billions of dollars for many banks. The estimate relies heavily on data including economic forecasting and historical loan outcomes, and has been described as part art, part science.

For Gullette, the big questions remain what level of detail will auditors require and how they will challenge the assumptions and economic models of their clients, he said.

Models Under the Microscope

Auditors say banks shouldn’t expect anything different even though their new rules place extra attention on the risks related to assumptions and judgments made by the company and enhance oversight of specialists, whose work often underpins estimates like CECL.

“We’ve always been auditing subjective numbers, so it’s not a new thing, necessarily, when there’s a lot of judgment. It’s going to come down to making sure you have your assertions corroborated,” said Reza Van Roosmalen, financial instruments accounting change leader with KPMG LLP.

Public companies should expect a lot of scrutiny on their economic models. The data feeding those models and the controls governing that data will all be top of mind, said Catherine Ide, senior managing director of professional practice and member services at the Center for Audit Quality.

Companies should be prepared to document their decisions and the steps they took to apply the new rules. They may need to consider changes to their internal controls especially as it relates to the data they are relying on, Ide said.

Three Rules Collide

Deloitte LLP has been focused on training its auditors, not just on CECL, but also on the audit rules for estimates and oversight of the work of specialists, said Khalid Shah, a partner in Deloitte’s national office who is leading the firm’s audit approach to CECL.

Auditors need to understand the basics of all three rules, and how to think through the challenges they will face, Shah said.

He rattled off a long to-do list.

In addition to new and changing controls, auditors need to be comfortable with the choices their client made in selecting data and how the methods are used to reach an estimate. They need to consider the source of the information and whether it is relevant and complete. They will review both the design and logic of the model and check its results, he said.

Banks have to use the new accounting to calculate the new loan loss reserve, and auditors will be checking that they didn’t just adjust the existing reserve up or down, Shah said.

And they aren’t waiting until January to get started. Auditors at Deloitte have been monitoring the modeling and other testing that their clients have been running for as long as two years at this point, Shah said.

In addition, auditors have already scrutinized initial estimates banks have shared with investors to prepare them for just how much their credit loss reserves will increase or decrease starting next spring.

Auditors have to be comfortable with those figures that are already showing up in quarterly reports, Shah said.

More Data, More Risk

CECL is a data-intensive estimate. And many of Grant Thornton’s clients have taken the opportunity to improve their data management and processing as they prepare to roll out the new accounting, Dyer said.

That means replacing spreadsheets with more automated systems, resulting in better, more rigorous modeling. But more automation also requires more reliance on testing internal controls around those systems and processes, Dyer said.

More data and the use of outside experts all exposes companies to more risks. And the new estimates standard offers auditors an important reminder not to trust the information they are provided blindly, said Kelly Richmond Pope, associate professor in the School of Accountancy and MIS at DePaul University.

The advent of cloud computing and big data—understanding who had access, what anomalies mean—puts more pressure on auditors. Those new tools and technology aren’t a replacement for healthy skepticism, she said.

“In order for us to help alleviate risk, we need to think more about the potential of fraud,” Richmond Pope said. “We do have a responsibility to the public to think about this more.”

—Nicola White contributed to this report.