Higher order risk preferences play an important role in economics, most prominently in the theory of saving under uncertainty. Leland (1968) suggests that under uncertain income prudent individuals increase savings as a precautionary measure. To test this proposition we present a new experimental method to elicit higher order risk preferences. The method we propose uses a non-parametric estimation of the utility function using P-splines. Using this method we can compute well-known theoretically derived measures of the intensities of prudence and risk aversion. We find comparable results to earlier studies with respect to classification of individuals as prudent or imprudent among a sample of poor households in Bogota. In addition, the results strongly support the theoretical prediction that uncertainty leads to increases in savings for prudent individuals. This suggests that this population group lacks alternative options to smooth consumption.