Virginia Postrel's article reminded me to read Conspicuous Consumption and Race by Charles, Hurst, and Roussanov. I was drawn to this article mainly because of some preconceived notions that this was true and the paper confirms that blacks and hispanics engage in higher levels of conspicuous consumption. Instead of making a race based argument they instead show that the higher spending is a result of the lower average income of their reference group (in this case blacks/hispanics in the state of residence). Unfortunately, this is a tautalogical argument since compared to whites the average state income for blacks and hispanics are expected to be lower so there is no way to separate the two hypotheses. It would be nice to have been able to do this at a lower level of aggregation - county level for instance but the data does not allow it.
However, in support of their hypothesis they also do different cuts within blacks/hispanics by age for instance and find that conspicuous consumption is lower the higher the average income of these subgroups. (No effects for education interestingly enough - see their Table 2.) So in some sense, they do try to tease out these effects but not sufficiently in my opinion.
They also claim to use a nationally representative data set of households but as far as I can tell it cannot possibly be representative because they were not able to perform separate analysis for Asians. (If the sample were representative, they should be able to do this.) I suspect that the deletions that were done on the same were responsible for this. Also, as far as I can tell the analysis does not make use of the survey weights so the results really are not representative.