|
Clementine
An improved risk management framework for tax collection
Lyne Sincennes
Director
Canada Revenue Agency
Canada
With personal income tax accounting for 47 percent of total budgetary revenues, efficient tax collection is crucial to support the Canadian government’s spending plans. Playing an important role in the process is the Canada Revenue Agency’s (CRA) Debt Management Branch, which is responsible for ensuring compliance with filing and reporting laws within a self-assessment tax system.
This session described how the CRA used SPSS’ Clementine workbench, to develop risk models to:
- Identify those non-compliant taxpayers who are likely to respond positively to appropriate enforcement actions
- Predict the likelihood of a taxpayer voluntarily resolving a tax debt
- Predict the probability that a collections officer will need to take enforcement action to resolve a debt
- Predict the likelihood of success of automated or call centre responses to debt issues.
In addition, CRA showed how they used SPSS models to identify and claim $7 million in new revenues from previously written-off, non-compliant taxpayers in a pilot project. Once in full production, the CRA expects to generate $180 million of additional revenue.
Back
|
Clementine, Text Mining for Clementine
Credit scoring in real time: reducing risk with text mining
Piero Biagi
General Director
Nolé SpA
Italy
Nolé SpA, controlled by Banca Agrileasing, is one of the leading Italian companies in the growing field of hiring innovative, all-inclusive value-added equipment solutions. Its clients include business and professional customers in the telecommunication, information technology, copying and digital printing, medical and banking sectors. Nolé operates in partnership with a group of vendors who supply the equipment offered for rental.
In particular, Nolé is the market leader in ‘small ticket’ rentals or leasing operations valued up to €25.000.
One of the biggest challenges facing rental businesses is assessing client credit risk, and in this session Nolé described how it used data mining technology to build an efficient real-time credit scoring system, and how the addition of text mining provided better predictions. They also presented results – a 55 percent increase in new business with no additional staff, and early defaults reduced from 3,6 percent to 1,5 percent – that demonstrate the effectiveness of their approach.
Back
|
|
SPSS, SPSS Server, Clementine
Data mining catches fraudulent patient sharing schemes
Marcee Sturino
Director, Data Informatics
TrustSolutions LLC
USA
TrustSolutions LLC is a Program Safeguard Contractor for the United States Department of Health and Human Services’ Centers for Medicare and Medicaid Services. It works with the U.S. federal government to detect and prevent fraud, waste and abuse in one of the country’s major health-care programs.
One particular issue that TrustSolutions was called upon to investigate was an allegation that health care facilities were sharing patients among themselves – and collecting fraudulent payments for each case.
In this session, TrustSolutions described how they used statistics to extract and format patient and claims data, and applied data mining to find and validate instances of fraud. As result, they disclosed, there were several referrals to law enforcement agencies in relation to potential overpayments exceeding US $14 million.
Back
|
SPSS, SPSS Server, Clementine
Using SPSS to validate Basel II requirements
Giacomo Petrini
Head of Models and Process Validation Services
UBI Banca
Italy
Under the first pillar of the Basel II Accords (International Convergence of Capital Measurement and Capital Standards), banks are required to assess how much capital they need as a safeguard against various types of financial and operational risk – including credit risk.
Banks must also establish an independent internal team to validate the methods used to perform the risk assessment – a very complex and time-consuming process.
This session was about how UBI Banca, a regional Italian bank with total assets of €119 billion, used statistics and data mining to simplify and speed up validation of its implementation of the Internal Ratings-Based approach to assessing credit risk.
The presentation focused on UBI Banca’s use of predictive analytics to validate internal estimates for credit risk and to verify the IT architecture of the internal rating system, making it easier to register the validation process with Bank of Italy in a relatively short time.
Back
|
|
Clementine
Using data mining to target defaulters and increase tax payments
Véronique Vandamme
Data Warehouse Manager
Elly Goossens
Business analyst
FOD Financiën
Belgium
According to the American inventor and statesman Benjamin Franklin, “In this world nothing can be said to be certain, except death and taxes”. Not everyone believes him, however, and some people will try hard to avoid or default on their tax obligations. But thanks to analytics, avoiders and defaulters in Belgium are finding it very difficult to escape.
The Minister of Finance instructed the Federal Public Service Finance (FPSF) to recover €75 million in unpaid taxes, and as part of the plan to achieve this, the organisation is building a warehouse of information about people who don’t pay their taxes.
This session was about how statistics are used to increase tax revenues by analysing a combination of internal and external sources, and how the expertise of FPSF’s analysts is being distributed in a format that can be used throughout the organisation.
The FPSF also confirmed results that exceeded expectations, with €4 million recovered in the first phase of trials and a further €2,7 million in the second. Perhaps Ben Franklin was right after all.
Back
|
SPSS Server, Clementine
Switching to automatic: how to assess credit risk in under five seconds
Edgars Peics
CRM IT Development Manager
Hansabank
Latvia
Latvia’s Hansabank Group, one of the region’s leading financial institutions, is preparing to transform its credit and pricing decision-making processes, moving from a manual procedure that takes 15 minutes to a fully automated, real-time system capable of arriving at a conclusion in between one and five seconds. The new system will comply fully with the Basel II requirements for ensuring that sufficient capital is reserved to cover possible credit defaults.
This session explained how Hansabank used SPSS Clementine to create dozens of risk monitoring and assessment models, which were then submitted to the national financial supervisory board for approval. When deployed, the models will produce credit risk scores in real time for 150.000 customers every month.
Other topics included how Hansabank met the challenge of aligning business and IT objectives; the lessons learned and the tools and practices that worked; and the bank’s vision for using modelling techniques to create business value.
Back
|