Legal compliance is an important part of certifying the correct behaviour of a business process. To be compliant, organizations might hard-wire regulations into processes, limiting the discretion that workers have when choosing what activities should be executed in a case. Worse, hard-wired compliant processes are difficult to change when laws change, and this occurs very often. This paper proposes a model-driven approach to process compliance and combines a) reference models from laws, and b) business process models. Both reference and process models are expressed in a declarative process language, The Dynamic Condition Response (DCR) graphs. They are subject to testing and verification, allowing law practitioners to check consistency against the intent of the law. Compliance checking is a combination of alignments between events in laws and events in a process model. In this way, a reference model can be used to check different process variants. Moreover, changes in the reference model due to law changes do not necessarily invalidate existing processes, allowing their reuse and adaptation. We exemplify the framework via the alignment of laws and business rules and a real contract change management process, Finally, we show how compliance checking for declarative processes is decidable, and provide a polynomial time approximation that contrasts NP complexity algorithms used in compliance checking for imperative business processes. All-together, this paper presents technical and methodological steps that are being used by legal practitioners in municipal governments in their efforts towards digitalization of work practices in the public sector.