An Adversarial Approach to Structural Estimation

Kaji, Tetsuya; Manresa, Elena; Pouliot, Guillaume

doi:10.2139/ssrn.3706365

Cited by 5 publications

(3 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This section states the asymptotic properties of our estimator. For more general results, we refer the reader to the earlier version in February 2022, Kaji, Manresa, and Pouliot (2022). Also, the theoretical results on the discriminator is given in Section S.2.…”

Section: Statistical Propertiesmentioning

confidence: 99%

“…Interestingly, our framework establishes a bridge between SMM and MLE. When we use a logistic discriminator with inputs equal to moments, the resulting estimator is asymptotically equivalent to optimally‐weighted SMM (Kaji, Manresa, and Pouliot (2023, Section S.1)). When we use the oracle discriminator, the resulting estimator is equivalent to MLE when the simulation sample size increases faster than the actual sample size, since our classification accuracy is a symmetrized Kullback–Leibler divergence.…”

Section: Introductionmentioning

confidence: 99%

“…1010 All theoretical asymptotic standard deviations of the adversarial estimators in Section 3 are calculated with a general version of Theorem 3 that allows

{normallim}_{n \to normal∞} n false/ m > 0

(Kaji, Manresa, and Pouliot (2022, Theorem 3)). …”

mentioning

confidence: 99%

See 2 more Smart Citations

An Adversarial Approach to Structural Estimation

Kaji,

Manresa,

Pouliot

2023

ECTA

Self Cite

View full text Add to dashboard Cite

We propose a new simulation‐based estimation method, adversarial estimation, for structural models. The estimator is formulated as the solution to a minimax problem between a generator (which generates simulated observations using the structural model) and a discriminator (which classifies whether an observation is simulated). The discriminator maximizes the accuracy of its classification while the generator minimizes it. We show that, with a sufficiently rich discriminator, the adversarial estimator attains parametric efficiency under correct specification and the parametric rate under misspecification. We advocate the use of a neural network as a discriminator that can exploit adaptivity properties and attain fast rates of convergence.

show abstract

Section: Statistical Propertiesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%