The Evolution of Irish Loan-Level Data Workshop on Integrated Management of Micro-databases 20 June 2013 Rory McElligott, Central Bank of Ireland The Roadmap 2013: A better Loan Level Database Central Credit Register 201X 2011: A Loan Level database 2010: Zero 2 1 Initial Motivation for Project • November 2010: Ireland agreed to a EU-IMF financial support programme. • Requirement to complete a bottom-up bank capital assessment by March 2011. • The review was completed in just over four months included loan-level data collection & quality remediation work, and loan-loss estimates • Exceptionally quick due to unique set of incentives • Review identified stress-case loan losses of €27.7bn 3 • The first collection was successful …but the process an expensive and unsustainable solution • Solution was to developed the internal IT infrastructure, skillsets and governance • Knowledge transfer from external consultants • Necessary preparatory work for future CCR project • Expanded and harmonised definitions, data fields and templates • Four data drops now completed 4 2 € Bn Loans to Irish residents Challenges remain 100 Frequency Semi-annual Timeliness 4-6 month lag Population coverage 3 main domestic banks (all offices) Covered Not Covered 80 60 40 20 Unstandardized Fragmented delivery files, definitions etc; (work in progress!) 0 5 How will the data be used? Late Stage Mortgage Arrears Focus is on informing and designing policy Analysts quickly embraced the new dataset Data: European ABS Outlook 2013, Deutsche Bank; Note this is not official data so please be cautious. 6 3 Early papers focused on understanding the condition of the Irish banking books December 2010 No Negative Equity Yes No 61.3% 33.9% Yes 2.4% 2.4% Arears (90+ days) 7 Source: McGuinness (2011), based on 31 Dec 2010 data. Analysts then began to model potential future outcomes 1-Year Transition Matrix (unconditional) Migration model uses historical performance data to estimate a detailed transition matrix through which the probability default of any loan can be estimated (Kelly 2012) 8 4 And are now working to inform broader economic policy • Assessing fiscal multiplier under strong sovereignbank inter-linkages • €2bn fiscal stimulus results in saving in bank capital of €660 million • Previous research indicated the stimulus would have a net cost of €1.3bn excluding banking capital gains – No allowance for other portfolios such as SME 9 Use of Loan-Level data for Statistics • It is not currently possible to get ‘good enough’ fit to statistical definitions/concept • Confidence is growing and the benefits of granular data are widely acknowledged • Statistics Division are concentrating on working with CCR project mangers to address major hurdles 10 5 Moving towards a Central Credit Register • The establishment is being fundamentally driven by supervisory needs • Motivation to improve credit market intelligence for lenders, borrowers, supervisors • Banking crisis exposed bad credit decisions • Goal that Credit Register will be operational by end2013. – phasing roll-out of CCR roll-out completed by 2016 11 CCR Project – Legislation Key aspects of the Bill include defining the:–Credit information providers • credit institutions and other regulated financial service providers, local authorities, NAMA. –Credit information subjects • individuals, corporations, sole-traders, partnerships and others threshold for reporting loans of at least €500. –Threshold for reporting loans of at least €500 –Levels of access to data by stakeholders 12 6 Summary Policy outputs justify the investment to date • Senior policymaker buy-in for flexible microdata Flexible Data + Research • These early analytical papers help maintain the momentum and support the business case to further develop the project. 13 Thank You Rory McElligott, Senior Economist, Central Bank of Ireland Email: rory.mcelligott@centralbank.ie Tel: +353 (0)1 2246626 14 7