Exploitation of vulnerable groups such as refugees for cheaplabour in agriculture and in the construction sector is a notorious phe-nomenon in Turkey. Up to 2017, only 1.3% of the around 3 mn Syrianrefugees registered in Turkey have been granted a work permit, leavingthe overwhelming majority dependent on undeclared employment withall its negative implications: high-risk jobs, pay below minimum wage,lack of access to social security. Mobile phone metadata allow for a de-tailed view on commuting routines and migration, possibly unearthingemployment situations which are not captured otherwise. This study pro-poses a methodological framework for identifying potentially undeclaredemployment among refugees in Turkey within the current situation. Todo so, it includes an early proof-of-concept based on a Difference-in-Differences approach by analyzing seasonal migration and commutingpatterns in two specific cases: during the late-summer hazelnut harvestin the province of Ordu and at the construction site of the Istanbul GrandAirport. The study finds clear indication for work-related migration andcommuting patterns among refugees hinting at undeclared employment.The proposed framework therefore provides an analytical instrument tomake targeted interventions such as controls more effective by detectingsmall areas where undeclared work likely takes place.