Agent-based models provide a promising new tool in macroeconomic research. Questions have been raised, however, regarding the validity of such models. A methodology of macroeconomic agent-based model (MABM) validation, that provides a deeper understanding of validation practices, is required. This paper takes steps towards such a methodology by connecting three elements. First, is a foundation of model validation in general. Second is a classification of models dependent on how the model is validated. An important distinction in this classification is the difference between mechanism and target validation. Third, is a framework that revolves around the relationship between the structure of models of complex systems with emergent properties and validation in practice. Important in this framework is to consider MABMs as modelling multiple non-trivial levels. Connecting these three elements provides us with a methodology of the validation of MABMs and allows us to come to the following conclusions regarding MABM validation. First, in MABMs, mechanisms at a lower level are distinct from, but provide input to higher levels of mechanisms. Since mechanisms at different levels are validated in different ways we can come to a specific characterization of MABMs within the model classification framework. Second, because the mechanisms of MABMs are validated in a direct way at the level of the agent, MABMs can be seen as a move towards a more realist approach to modelling compared to DSGE.

The main aim of this dissertation is to contribute to the systematic understanding of modeling practices within macroeconomics. As its chosen method, it analyzes three particular cases, which, each in their own way, may be taken to be representative of modeling practices today. Subsequently, the results from the case studies are integrated and evaluated. The first case is discussed in Chapter 2 and focuses on the empirical validation of macroeconomic agent-based models. It is shown that agent-based models in macroeconomics can be best understood by considering them as complex systems with a multitude of interaction levels. Empirical validation tests, as observed in practice, can be related to these levels of interaction. Furthermore, I distinguish between validation tests directed at the model target and the model structure and consider how these apply to macroeconomic agent-based models. The second case is discussed in Chapter 3 and concentrates on the shift in dynamic stochastic general equilibrium (DSGE) models from being calibrated to being estimated. Estimation is considered preferable to calibration for several reasons, in particular because it allows for the parametrization of models with a large number of parameters. In turn, this makes possible the construction of larger models that incorporate more of the complexity of the real-world economic system. This shift required DSGE models to evolve into what I label as a ‘hybrid model structure’, which is a model with a representational core supplemented with non-representational stochastic elements. This hybrid model structure has come under fire from several authors. I argue that this critique is warranted if it is understood as a disruption in the relationship between the outcome of an empirical validation test and the validity of the model. The third case, presented in Chapter 4 and discusses interdomain model transfer. The case studied is that of the Yule process, a model originally developed in evolutionary biology but later reused as a model for firm growth. The aim of the chapter is to provide an explanation of why such model transfer may appear. It presents a framework of model transfer in which the various validation criteria distinguished in Chapter 3 play a central role. A model is transferred from an original to a new domain when it is found to be useful in both domains. This is the case when overlap is found between the validation criteria of both domains. Special attention is paid here to overlap between phenomenological criteria, which is enabled through the existence of observed universal patterns. Overlap of this type is an important explanation in the case of the Yule process. The main result of this dissertation is the model construction framework presented in Chapter 5. In this framework, the insights of the case studies are integrated. The framework is centered on the concepts of model purpose, invariance, model validation, and model scope, as well as how these concepts fit together. Ultimately, this is a framework that provides a systematic understanding of macroeconomic modeling practice. .

Model transfer refers to the observation that particular model structures are used across multiple distinct scientific domains. This paper puts forward an account to explain the inter-domain transfer of model structures. Central in the account is the role of validation criteria in determining whether a model is considered to be useful by practitioners. Validation criteria are points of reference to which model correctness for a particular purpose is assessed. I argue that validation criteria can be categorized as being mathematical, theoretical or phenomenological in nature. Model transfer is explained by overlap in validation criteria between scientific domains. Particular emphasis is placed on overlap between phenomenological criteria. Overlap in phenomenological criteria can be explained through the notion of universal patterns. Universal patterns are abstract structures that can be made to refer to multiple distinct phenomena when coupled with phenomena-specific empirical content. I present the case study of the Yule Process, in which universal patterns play a crucial role in explaining model transfer. This paper provides an account of model transfer that stays close to modelling practice and expands existing accounts by introducing the notion of universal patterns.