Digitalization, paired with AI helps labour inspectorates address undeclared work in Albania

Representative image of Artificial Intelligence by Tiago Louvize on unsplash

Reporting and inspection planning have become more efficient for Albanian labour inspectors with the new data management and risk assessment tool developed by the ILO with EU funding. It also improved the detection of undeclared work. As the first results of the ILO-led pilot show, digitalization of workflows and systematic data collection, paired with artificial intelligence to analyse data, can play a crucial role in improving labour inspectorates’ effectiveness to address undeclared work.

In 2021, the ILO and the Albanian Inspectorate of Labour and Social Services (SILSS) embarked on a joint effort to bolster the inspectorate’s effectiveness in addressing undeclared work. Three years down the road, the capacities of the SILSS have become stronger, thanks to the tools and the roadmap for digitalization of labour inspectorates developed by the ILO project “Employment and Social Affairs Platform phase two (ESAP2)”.

The labour inspectorate in Albania, similarly to other public sector institutions, is perceived as lagging behind the private sector in technological capacities. The SILSS management had a strong political will to reduce undeclared work and improve its overall institutional performance. Data analytics, digitalization and advanced technology were seen as enablers to achieve these objectives. Despite scarce resources, the SILSS digitalized its processes and used the data collected to build a red-flags risk assessment tool for labour rights violations falling under its authority. It was also imperative to upgrade the outdated technology and infrastructure of the SILSS, and upskill labour inspectors in the use of data-driven inspection planning.

The ILO with its technical expertise supported the SILSS to operationalize its goals and develop advanced electronic case and data management and risk assessment tools. The power of artificial intelligence was leveraged to improve planning of labour inspections.

Between 2021 and 2023, a substantial overhaul of the SILSS took place. The new Matrix of Inspections and Risk Assessment (MIRA) provides a case and data management tool that labour inspectors use in their daily work jointly with a risk assessment tool that uses data mining and machine learning (DM&ML), a form of artificial intelligence.

Labour inspectors benefit from a case management tool offering easy access to labour legislation and regulations; a more efficient and clearly defined workflow resulting in increased quality and efficiency of inspection visits; as well as better structured and standardized data allowing for higher quality and reliability of inspection data. The risk assessment tool utilizes historical data to predict potential violators and hotspots of undeclared work or other labour law and Occupational Safety and Health violations. Furthermore, it enables SILSS to deploy resources effectively to prevent or uncover undeclared work and increase employment formalization. In addition, DM&ML helps the SILSS to spot labour law violation trends early, providing it with a greater capability to prevent new trends when they are in their embryonic phase.

According to an ILO study conducted as part of the ESAP2 project, the use of the risk assessment tool utilizing DM&ML increases the accuracy of prediction of undeclared and underdeclared work by 30 per cent compared to the red-flags risk assessment used by the SILSS prior to MIRA.

The predictability power of this risk assessment tool is expected to increase as the datapoints in the dataset increase. However, to ensure that bias in training data is reduced and new trends in employer behaviour are identified, the SILSS is making sure that around 30 per cent of inspection visits are not planned based on DM&ML modelling and predictions, but randomly or due to other grounds.

Rolling out a DM&ML risk assessment system implies several challenges, such as staff resistance, project delays and failures. To address potential challenges, the SILSS Management Team actively participated in its implementation of the tool and training 137 inspectors (out of which 71 women) of the Central and Regional Directorates. This ensured full ownership and upskilling of users prior to adoption of the new tool. the lead in working with software development and in training 137 inspectors (out of which 71 women) of the Central and Regional Directorates. This ensured full ownership and upskilling of users prior to adoption of the new tool.

The MIRA tool went live in November 2023, and has been working as intended, enabling labour inspection in Albania to better enforce the labour regulatory framework thus guaranteeing the effective protection of Albanian workers.