Every quarter, Google’s accounting team had been downloading excel-based gain/loss reports from State Street custody accounting, filtering and selecting items for more detailed review. Liaising with the treasury risk and portfolio teams, the company’s accounting department then followed up with external managers and their own research to manually identify credit events. This process relied on a relatively arbitrary review threshold and the ability of the various treasury teams to appropriately identify credit-related events. It also depended on accounting and risk professionals to successfully evaluate the impact of those events on observed declines in fair value. These dependencies were costly, not only in terms of time and effort, but also with regard to risk or error and lack of visibility.

In Q1 2012, Google implemented an issuer level credit modelling framework, custom developed by doctorate researchers at Moody’s Credit Research, to enhance the precision, quality, and efficiency of its effort-intensive asset impairments process. “The model consumes Google investment accounting data from our custodial agent, State Street Trust Company, on a monthly and quarterly basis. Data is transmitted directly via and from Secure File Transfer Protocol (SFTP). Once a month, this data is uploaded and integrated with the ‘Moody’s Credit Universe’ data suite within Moody’s’ proprietary CreditEdge+ analytics application,” says Joe Stanfill, Analyst at Google.

A newly merged, credit data enriched file is then processed through the framework for all holdings within Google’s emerging market, investment grade, and high yield funds groups. Together, these funds represent around 9,000 holdings and $6 billion assets under management (AUM) as of April 2012. The resulting output files are delivered by Moody’s one day after they receive each accounting data ‘input’ file and provide a comprehensive view of all potential impairment candidates in eligible portfolios. Three distinct measurements (or credit ‘perspectives’) included in these files each reflect different assumptions about the cause, severity and probability of persistence of known credit events which have impacted each issuer during any given period.

“Data retrieval, research, identification, and evaluation efforts are now built into the underlying data, modelling, and framework solution itself, eliminating the need for arbitrary thresholds and subjective judgements of causation and impact,” Stanfill explains. All declines in fair value in eligible portfolios are systematically reviewed, instead of by limited sampling. The model’s credit metrics identify, with much greater precision, which holdings’ losses are more likely due to credit events and provide guidance as to their possible persistence and severity.

The re-engineered asset impairments process is fully automated, excluding the generation of impairment accounting entries, and saves Google treasury up to 9 -12 man hours per quarter end (potentially more in volatile markets). According to Stanfill, “It also provides enhanced visibility with regular reporting and the additional option to run ad-hoc reports on request, thereby reducing intra-quarter risk of unobserved credit impacts to portfolio value.” Furthermore, the solution allows comprehensive coverage of all holdings in Google’s non-agency-backed portfolio groups. Google’s accounting team, once able only to review perhaps 200-300 securities, with loss exceeding a certain threshold at quarter end, is now able to review on average 2000-3000 securities – in two thirds of the time.

To Google’s knowledge, this solution is the first of its kind in the corporate treasury space. Benchmarking against six corporate treasuries in the technology industry with equivalent cash management profiles, Google determined that all were (also) utilising highly-manual, subjective review processes driven by Microsoft Excel and professional judgment. This new framework integrates high-quality credit data with sophisticated, yet transparent, modelling methodology and technology to not only automate the process, but significantly enhance overall precision.