Data Quality for the Economic and Financial Statistics Collection
Australian Prudential Regulation Authority (APRA) has released a Reporting Practice Guide, RPG 702.0 ABS/RBA Data Quality for the EFS Collection (RPG 702.0), which provides guidance on managing data quality for entities reporting under the economic and financial statistics (EFS) data collection. RPG 702.0 provides benchmarks for identifying reporting errors and mandates all Authorised Deposit-taking Institutions (ADIs) and Registered Financial Corporations (RFCs) that are required to report under the EFS reporting standards to have in place systems, processes and controls to assure the reliability of reported information. The regulated entities are expected to design and implement control throughout the data life-cycle – including, but not limited to, data capture, processing, retention, preparation and submission of reports. The instructions provided in RPG 702.0 are in alignment with the guidance provided under CPG 235, the prudential practice guide for managing data risk.
CPG 235 – Managing Data Risk
CPG 235 defines ‘Data’ as “the representation of facts, figures and ideas”. It defines ‘Data Risk’ as a subset of operational risk encompassing the risk of loss resulting from inadequate or failed internal processes, people and systems or from external events impacting data. Hence, the regulated entities must manage data risk in alignment with their operational risk framework and consider the end-to- end use of data and related control environment. CPG 235 also recommends the assessment and management of data quality as a useful technique for managing data risk. It provides an illustrative list of data quality dimensions and risks to be considered at all stages of data life-cycle. ADIs and RFCs which are required to adhere to RPG 702.0 can adhere to this data risk management framework.
Exhibit 1: CPG 235 Data Risk Management Framework
RPF 702.0 – Benchmarks for reporting data errors
The Reserve Bank of Australia (RBA) and the Australian Bureau of Statistics (ABS) use the data collected through EFS for formulating policies which affect Australian economy. Hence, RBA, ABS and APRA (collectively referred as agencies) consider it extremely important to ensure the correctness of data collected. The agencies have provided benchmarks which define the magnitude of data errors that they are willing to tolerate for different reported fields across EFS reporting forms. The reported fields have been classified into three groups – Very High, High and Standard, in the decreasing order of priority for data quality. These priorities indicate the importance given to the reported fields by the agencies based on their impact on policy decisions. And predictably, the defined tolerance for errors decrease with increase in priority.
Exhibit 2: Classification of reported fields across priorities
The agencies define a separate set of benchmarks for large institutions – defined as an ADI or an RFC with greater than or equal to A$200 billion in total assets measured on a domestic books basis. The tolerable margins for errors are relatively narrow for large institutions as the absolute values of reported fields by larger institutions are large enough to significantly affect the policy decisions by the agencies. Separate sets of benchmarks are provided for stock, flow and rate data types, with relatively relaxed tolerance levels for flow data type as they exhibit a natural period-to-period volatility. Besides, RPG 702.0 defines benchmarks, both in terms of absolute dollar figures and percentage deviation from the actual values – for reported fields whose values are closer to zero, the agencies check only the absolute dollar figures as the percentage values could be high due to low base effect.
Exhibit 3: Benchmarks for identifying reporting errors for a large institution
Exhibit 4: Benchmarks for identifying reporting errors for entities which are not large institutions
Exhibit 5: Benchmarks for identifying reporting errors in ‘rate’ data types for large institutions
Exhibit 6: Benchmarks for identifying reporting errors for ‘rate’ data types for entities which are not large institutions