Evaluation models for housing loan allocation in the context of floats

In this paper, we analyze the impact of contextual changes to the size, structure and behavior of a qualitative multi-attribute model. The study involves an evaluation model for allocating housing loans in the Housing Fund of the Republic of Slovenia. So far, this model was used in 13 completed floats of loans for citizens. As a result of considerable contextual changes that occurred from float to float, the model had to be adapted accordingly. These changes were reflected in the modifications of model structure, used attributes, and decision rules at various levels of the model. Based on the analysis of time series of different model parameters, we articulate some guidelines for efficient model management, and justify the flexibility of the approach and its robustness to contextual changes. In particular, we show that a model can be designed so that relatively large changes in the decision environment cause minor and isolated changes in the model.