In the Groningen province, situated in the northern part of the Netherlands, a large natural gas field was discovered in 1959 and in 1964 the gas extraction started. Due to the gas extraction in the Groningen area, undesired induced earthquakes are generated. The strongest earthquake registered to date was in 2012, 3.6 of the Richter scale. Not being designed and constructed to sustain seismic loading, masonry buildings in the area show higher vulnerability than other more recent structures. In 2013 extensive research on the seismic risk has been initiated in order to assess structure and infrastructure vulnerability to seismic events. Smeared plasticity shell element models have shown to achieve very accurate results in blind predictions of full-scale laboratory tests. However, they are computationally very demanding and hence not suitable for parametric studies and fragility curves development. Alternatively, a computationally efficient lumped plasticity model is proposed, meant to be representative of the in-plane shear and rocking behaviour of URM wall piers. The macro-element model is calibrated with quasi-static tests performed in laboratories at TU Delft and Eucentre. In addition, smeared plasticity numerical simulations are used for the calibration of the energy-based phenomenological parameter in the flag-shaped rule for rocking behaviour. A new hysteretic rule has been developed for modelling the flange effect that characterizes slender piers of the Groningen URM buildings. It provides the abstraction for a one-dimensional rate-independent flag-shaped hysteresis with asymmetric backbone. Non-linear time history analyses are performed to assess the response of wall components and to validate the approach against the smeared-plasticity model. Comparable results are achieved in terms of collapse-mechanism prediction, base shear forces, maximum drifts and energy dissipation. Finally, the macro-element is assembled in an equivalent-frame model which represents a prototype house tested quasistatically at TU Delft. The blind prediction displacement-controlled test shows a 20%capacity overestimation from the lumped plasticity model, proving the necessity of parametric studies for the capacity estimation improvements. NLTH analyses lead to comparable results between the two approaches, in terms of failure mechanism prediction and hysteretic behaviour. The analyses performed with the proposed approach turned out to be faster by two orders of magnitude compared to the shell element models.