This paper reports on an analysis of life satisfaction data collected as part of the first wave of the
Household, Income and Labour Dynamics in Australia (HILDA) Survey. More specifically, the
clustered nature of the HILDA sample was used to test the role of neighbourhood effects in
accounting for inter-personal differences in self-reported life satisfaction scores.
A regression model predicting individual differences in life satisfaction was developed and tested
for men and women separately. When this model was estimated allowing for fixed
neighbourhood effects (based on the Census Collection District in which a sample member
resides), strong support for sizeable effects were found. Indeed, observable individual and
household characteristics (such as age, sex, employment status and household income) were only
found to account for about 12 to 14 per cent of the variation in measured life satisfaction. Of the
variance unexplained, close to 10 per cent could be accounted for by unobserved differences
across neighbourhoods.
While identifying the presence and magnitude of neighbourhood effects proved to be relatively
straightforward, determining the source of these neighbourhood differences is a very different
matter. Essentially, these neighbourhood effects can arise either because individuals in the same
neighbourhood tend to behave similarly because they face similar environments or have similar
characteristics, or because the behaviour of individuals is affected by the behaviour of other
residents of the neighbourhood. Some evidence was uncovered to suggest that the latter type of
effect might be relatively more powerful in explaining differences in life satisfaction.
Unfortunately, this conclusion is tentative at best, with measurable neighbourhood characteristics
only found to have a relatively small impact on the overall explanatory power of the regression
models.