Elia Lapenta

PhD Candidate in Economics, Toulouse School of Economics.

I am currently on the job market and will be available for interviews at the 2019 European Job Market in Rotterdam and the 2020 ASSA annual meeting in San Diego.


Primary Research Field: Theoretical Econometrics , specifically

  • Semiparametric and nonparametric estimation and testing
  • Bootstrap methods

Secondary Research Fields: Empirical games of incomplete information, boosting methods, and high dimensional econometrics


CV Research Statement


Email: elia.lapenta(@)gmail.com

RESEARCH PAPERS


Title: A Bootstrap Specification Test for Semiparametric Models with Generated Covariates

Job Market Paper

Author: Elia Lapenta

Abstract: Models with nonparametrically generated variables -i.e. variables that are not observed but are nonparametrically identified- are common in Economics. Examples of these are models with endogenous regressors, where the endogeneity is handled by control functions; sample-selection models, where an outcome variable is observed only on the selected sample; or empirical games with incomplete information. In this paper, I provide a test to check the correct specification of these models. The statistic I propose is a weighted sum of the estimated residuals, and is easy to compute. By using tools from Empirical Process Theory, I show that its asymptotic null distribution is a transformation of a Gaussian process. To obtain the critical values, I develop a two-step Wild Bootstrap. The test involves bias corrections for the nonparametric estimators, so it can be implemented without undersmoothing. A Monte-Carlo simulation study shows the good small-sample performance of the test.


Title: Testing Bayesian-Nash Behaviors in Binary Games with Incomplete Information and Correlated Types

Authors: Elia Lapenta and Pascal Lavergne

In this paper, we provide a test to check if the distribution of the observed data can be rationalized by a unique Bayesian-Nash equilibrium of a binary game with incomplete information, where agents’ types are allowed to be mutually correlated. Testing such a kind of assumption is useful for two reasons. First, the uniqueness of the Bayesian-Nash equilibrium is key to identify the fundamentals of the game. Second, testing the Bayesian-Nash behavior is interesting per se, as it is an economic assumption often postulated in game theoretical models. The test we propose relies on rationalization results in Liu et al. (2017). In order to construct our test statistic, we implement an L2-boosting procedure from the Machine Learning literature. This is quite effective to control the bias arising in our context. We derive the asymptotic behavior of the test statistic, and propose two bootstrap procedures to obtain the critical value. A Monte Carlo experiment shows the good small-sample performance of the test.


Title: A Nonparametric Encompassing Test Based on L2 Boosting

Authors: Elia Lapenta and Pascal Lavergne

According to the Encompassing principle, a model M1 encompasses a model M2 if the former is able to explain the results of the latter. The encompassing tests so far provided either rely on parametric functional forms or, when relying on nonparametric specifications, they condition the analysis on fixed values of the explanatory variables. Accordingly, the results obtained can be considered as conditional on these specific features. In this paper, we provide a nonparametric encompassing test. Our procedure does not rely on either functional forms nor on specific values of the explanatory variables. We propose a statistic that is computed according to an L2 boosting algorithm. This procedure allows to obtain a good robustness with respect to the choice of the smoothing parameter. We propose to simulate the critical values by a Wild-Bootstrap procedure. In a Monte-Carlo simulation study, we show the attractive features of our test.

REFERENCES

Professor Pascal Lavergne

Professor Jean-Pierre Florens

Professor Juan Carlos Escanciano

Professor Ingrid Van Keilegom

CONFERENCE, WORKSHOP, AND SEMINAR PRESENTATIONS

2019: IAAE Conference (June), Nicosia; ENTER Seminar (September), Mannheim; ESEM (August), Manchester; Bristol-TSE Econometrics Workshop (September), Toulouse; Econometrics Seminar (October), Nottingham; French Econometrics Conference (November), Marseille;

2018: International Society of Nonparametric Statistics (June), Salerno;

2017: TSE Econometric Workshop (March), Toulouse; ENTER Jamboree Conference (April), London;

2016: TSE Econometric Workshop (February), Toulouse; ENTER Jamboree Conference (April), Madrid; TSE PhD Students Workshop (June), Toulouse;

TEACHING

TA for Econometrics, Master 1 (in English) 2018-2019

TA for International Economics, Bachelor (in French) 2017-2018

TA for Econometrics, Master 1 (in English) 2016-2017

TA for Econometrics, Master 1 (in English) 2015-2016

ContactMail : elia.lapenta(@)gmail.com ; elia.lapenta(@)tse-fr.eu