Business Intelligence in Taxation

Posted: October 2006 C: Business Intelligence Posted by: K. Panayotakis

Government Finance Divisions manage the execution of the fiscal policy: • management of the taxation and tax audit process • management of the budget execution process

Within the taxation framework, each State aims to support its taxation policy: • Support the efficient capture and processing of tax declarations • Assure the collection of tax revenue according to plans • Manage and reduce the risk of Citizen non-compliance: reduce tax evasion levels and fight financial crime • Support the analysis of alternative future tax policies and relevant decision making, e.g. consequences of a change on State revenue • Enhance the Citizen experience, during the tax compliance process (e.g. offer e-government services).

It is proposed that a ‘tax monitoring data mart’, be designed based on the dimensional model approach. A proposed high–level logical data model is depicted in the tax monitoring data mart figure, based on the assumption that a critical mass of data (geographic & time dimension) in a sufficient data quality exists, rendering the implementation feasible.

The proposed high-level data model depicted in the figure, is a star schema with a number of fact tables surrounded by a set of common (or conformed) dimensions. The fact tables are of the ‘accumulating snapshot’ type, meaning that they accumulate information throughout each individual taxation transaction lifecycle. Most taxation types have a yearly lifecycle (i.e. personal income tax). VAT involves more frequent transactions.

This data modeling approach offers advantages, compared to the operational systems data models (usually normalized database schemas): • The model is quite simple and can be easily understood by business (non-IT) users • The model’s simplicity has also positive consequences to the performance of queries

Based on this infrastructure which would monitor the taxation lifecycle of various taxes, the Government can produce information: • on a taxpayer’s holistic view (all transactions she / he made) and support the non-compliance risk scoring and general tax audit process • on tax collection time-series analysis • on the time it takes between reception of tax declarations & clearance • on the tax collection speed • on the Taxpayer demographics (i.e. average declared income per professional code, geographic distribution of income) • total tax collected by tax office, compared to tax office attributes

The model is expandable when the following principle is applied: the star schema dimensions (time, tax office, taxpayer) are conformed allowing the application of queries via conformed dimensions across different fact tables (drill across queries). This approach is known as Datawarehouse Bus Architecture.

A tax audit case data mart is depicted in the following tax_audit_case_data_mart