1
$\begingroup$

Suppose I observe a random sample $X_1, ..., X_n$ from some continuous distribution $F$ and I want to estimate the parameter $\eta_q \equiv \mathbb{E}[g(q, X_i)]$ where $q$ is some specified quantile, e.g. the median, of the distribution of $X_i$ and $g(q, X_i) = \mathbb{1}(X_i > q) X_i$. If $q$ were known, then I could appeal to the central limit theorem to argue that $\sqrt{n}(\widehat{\eta}_q - \eta_q) \rightarrow_d N(0, \sigma^2_q)$ where $\widehat{\eta}_q = n^{-1}\sum_{i=1}^n g(q,X_i)$ and $\sigma^2_q = Var[g(q,X_i)]$.

But suppose that $q$ is not known and must be estimated by a corresponding sample quantile $\widehat{q}$. My question is how to account for the preliminary estimation of $q$ in the asymptotic variance of $\widehat{\eta}_{\widehat{q}}$. My confusion comes from the lack of smoothness of $g$ as a function of $q$. I understand how to handle the more standard case where a moment function depends smoothly on a preliminary estimator for some parameter $\theta$.

Update: It looks like this example should be covered by the approach described in Section 3.2 of Chapter 37 in the Handbook of Econometrics: "Empirical Process Methods in Econometrics," by Andrews (1994). Assuming this turns out to be correct, I will add details below.

$\endgroup$
2
  • $\begingroup$ Are you fixing the quantile (i.e say, so that we have $0.25$ probability to the right), or the corresponding value in the support of $X$? (say, we want $X>3$?). You write "$q$ is some specified quantile", but then you use the symbol $q$ as a value in the support of $X$, this is why I ask for clarification $\endgroup$ Aug 6, 2017 at 19:15
  • $\begingroup$ I'm fixing the quantile, say for instance the median: $F(q)=0.5$ $\endgroup$ Aug 6, 2017 at 23:12

1 Answer 1

0
$\begingroup$

This example is covered by the approach described in Section 3.2 of Chapter 37 in the Handbook of Econometrics: "Empirical Process Methods in Econometrics," by Andrews (1994). See also Newey and McFadden (1994) Section 7. An accessible introduction to the basic idea of how to proceed in non-smooth GMM / method of moment problems, focusing on the quantile case, is available here:

http://www.cemfi.es/~arellano/nonsmooth-gmm.pdf

When I have more time, I'll type up the full solution to the specific problem posed above.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.