In this thesis, I investigate how households’ portfolio allocations change in response to wealth fluctuations. The idea is based a paper by Brunnermeier and Nagel (2008) which used panel data from the USA to test implications of the difference-habit models of consumption. Models with difference-habits have succeeded in explaining many macroeconomic phenomena such as mean and countercyclicality of asset return risk premia, and stylized facts about asset returns and the business cycle. The aim of this project is to find evidence in micro data to support or reject the theory of habit preferences.

One of the implications of difference-habits is that households' relative risk-aversion should vary with wealth. Brunnermeier and Nagel conducted an econometric analysis to test whether changes in household wealth lead to 1) changes in the likelihood of households to participate in the stock market, and 2) changes in the asset-allocation of households between risky and risk-free assets. I replicate this study using Australian household data, including variables for household wealth, stock market participation and value of holdings, as well as demographic and household characteristics as control variables. I find that the attitudes to risk displayed in the data are not consistent with habit preferences, and are better described as constant relative risk aversion.