1: \begin{abstract}

2: This paper proposes a density-weighted average derivative estimator

3: based on two noisy measures of a latent regressor. Both measures have

4: classical errors with possibly asymmetric distributions. We show that

5: the proposed estimator achieves the root-$n$ rate of convergence,

6: and derive its asymptotic normal distribution for statistical inference.

7: Simulation studies demonstrate excellent small-sample performance

8: supporting the root-$n$ asymptotic normality. Based on the proposed

9: estimator, we construct a formal test on the sub-unity of the marginal

10: propensity to consume out of permanent income (MPCP) under a nonparametric

11: consumption model and a permanent-transitory model of income dynamics

12: with nonparametric distribution. Applying the test to four recent

13: waves of U.S. Panel Study of Income Dynamics (PSID), we reject the

14: null hypothesis of the unit MPCP in favor of a sub-unit MPCP, supporting

15: the buffer-stock model of saving. 

16: 

17: \vspace{.15in}

18: \noindent\textbf{Keywords:} Average derivative, latent variables, income dynamics, consumption \\

19: \noindent\textbf{JEL Codes:} C14, C23, D31

20: 

21: \end{abstract}

22: