Methods and Tools for the Microsimulation of Household Expenditure

Spending by households represents a significant component in the UK economy and the ability to model the effects of socio-economic change on household expenditure is crucial at both the commercial and governmental level. This is usually done by estimating the parameters of a demand system. However, there are several difficulties associated with this approach including, representing the heterogeneity of economic units, the dimensionality of a complex budget set and the specification of the functional form. One way to avoid these is to develop a model that directly simulates the individual units in what is known as a microsimulation. However, models of this type have been found to be complex and expensive to develop. This thesis investigates the possibility of simplifying the development process by using an agent-based modelling toolkit called NetLogo. The idea is tested by constructing a model to project the demographic characteristics of the UK population over time, showing that NetLogo provides a powerful and efficient platform for microsimulation modelling. Then it applies what is known as a random assignment scheme to model household expenditure. This is based on the idea of copying the expenditure pattern from a donor, which is in some sense similar to the receiving unit. Random assignment is then tested by developing a series of models of the effect of demographic and economic change on UK household expenditure patterns. The thesis contributes to methods and tools for modelling household expenditure by developing a framework for the analysis of household spending patterns based on the application of micro-level concepts and techniques throughout. This makes it possible to do what could not be done before which is to have a convenient way to model household expenditure that places no limit on the level of disaggregation or the number of goods represented.