Using data from the Consumer Expenditure Survey, this paper presents a
simple test that provides an explicit estimate of the parameter in th
e utility function that reflects the strength of the precautionary sav
ing motive, the coefficient of relative prudence. The test yields a fa
irly precise estimate of a small precautionary motive; in fact, the es
timate is too small to be consistent with widely accepted beliefs abou
t risk aversion. The presence of liquidity-constrained households does
not appear to explain this finding, and there is some evidence that s
elf-selection of households into risky environments also cannot explai
n the results.