Name in original language
Ábendingar á framtali
Initiative overview
The Icelandic Revenue and Customs tax control function reviews large numbers of tax forms each year to, among other things, identify undisclosed rental income, which is time-consuming and competes with other high-priority compliance work. To address this, we developed a predictive model that estimates the likelihood that a submitted tax form contains undisclosed rental income, allowing potential cases to be identified earlier and handled more efficiently.
The main objective is to improve the effectiveness and timeliness of compliance activities while using staff time and expertise where it creates the most value. By prioritising forms with a higher predicted risk, the initiative helps focus manual review on the most relevant cases, supports more consistent decision-making, and reduces the need for broad, resource-intensive screening.
The initiative has already contributed to substantial gains in tax collection and has freed up time and manpower that can be redirected to more complex tax control cases. Over time, we have moved from a primarily manual approach (where cases are found through broad review and experience-based selection) to a more data-driven prioritisation process that supports staff with an additional, objective signal.
Looking ahead, the model is intended to be further embedded into day-to-day workflows and governance so it becomes a standard part of how cases are selected for review. The approach can also be extended to other compliance areas where similar challenges exist (high volumes, limited capacity, and a need to identify higher-risk submissions early), subject to data availability, legal and privacy requirements, and ongoing monitoring to ensure the model remains accurate and fair.
Results, outcomes and impacts
Across 2022–2025, the model improved targeting of tax forms likely to include undisclosed rental income. It is used to prioritise around 3,500 cases per year, with an average confirmed-finding (hit) rate of ~86%. In total, the initiative has contributed to roughly ISK 5 billion (about EUR 33 million) in additional tax collected. Impacts were measured through back-testing on historical data and before/after operational comparisons, tracking tax outcomes and model performance over time. The initiative is also estimated to save ~450 person-months of manual screening effort per year, freeing capacity for more complex compliance work.
