Accumulated theoretical and empirical evidence suggests that wine prices de
pend on quality, reputation and objective characteristics. Unlike previous
studies, we recognize that quality and reputation are latent constructs and
therefore employ factor analysis and 2SLS techniques to consistently estim
ate hedonic prices in the presence of attributes measured with error. The a
pplication to Australian prmium wines points to significant reputation effe
cts but insignificant quality effects. It is also illustrated that inapprop
riately using standard OLS procedures can seriously distort the statistical
significance of attributes, the implicit marginal attribute prices, and th
e predictions of 'average' prices for a given set of characteristics.