• Corpus ID: 216553192

Structural Regularization

@article{Mao2020StructuralR,
  title={Structural Regularization},
  author={Jiaming Mao and Zhesheng Zheng},
  journal={ArXiv},
  year={2020},
  volume={abs/2004.12601}
}
We propose a novel method for modeling data by using structural models based on economic theory as regularizers for statistical models. We show that even if a structural model is misspecified, as long as it is informative about the data-generating mechanism, our method can outperform both the (misspecified) structural model and un-structural-regularized statistical models. Our method permits a Bayesian interpretation of theory as prior knowledge and can be used both for statistical prediction… 

Ensemble Learning with Statistical and Structural Models

TLDR
A set of novel methods for combining statistical and structural models for improved prediction and causal inference are proposed, including a doubly robustness property and a weighted ensemble that has the ability to outperform both models when they are both misspecified.

Machine learning and structural econometrics: contrasts and synergies

We contrast machine learning (ML) and structural econometrics (SE), focusing on areas where ML can advance the goals of SE. Our views have been informed and inspired by the contributions to this

References

SHOWING 1-10 OF 85 REFERENCES

Ensemble Learning with Statistical and Structural Models

TLDR
A set of novel methods for combining statistical and structural models for improved prediction and causal inference are proposed, including a doubly robustness property and a weighted ensemble that has the ability to outperform both models when they are both misspecified.

Minimizing Sensitivity to Model Misspecification

TLDR
This work proposes a framework for estimation and inference about the parameters of an economic model and predictions based on it, when the model may be misspecified using a local asymptotic approach and derives formulas to construct estimators whose mean squared error is minimax in a neighborhood of the reference model.

The Use of Structural Models in Econometrics

This paper discusses the role of structural economic models in empirical analysis and policy design. The central payoff of a structural econometric model is that it allows an empirical researcher to

Taking the Dogma out of Econometrics: Structural Modeling and Credible Inference

Without a doubt, there has been a "credibility revolution" in applied econometrics. One contributing development has been in the improvement and increased use in data analysis of "structural

Sufficient Statistics for Welfare Analysis: A Bridge between Structural and Reduced-Form Methods

The debate between "structural" and "reduced-form" approaches has generated substantial controversy in applied economics. This article reviews a recent literature in public economics that combines

Stable Prediction with Model Misspecification and Agnostic Distribution Shift

TLDR
A novel Decorrelated Weighting Regression algorithm which jointly optimizes a variable decorrelation regularizer and a weighted regression model to improve the accuracy of parameter estimation and stability of prediction with model misspecification and agnostic distribution shift.

Double machine learning for treatment and causal parameters

TLDR
The resulting method could be called a "double ML" method because it relies on estimating primary and auxiliary predictive models and achieves the fastest rates of convergence and exhibit robust good behavior with respect to a broader class of probability distributions than naive "single" ML estimators.

REGULARIZED OPTIMIZATION IN STATISTICAL LEARNING: A BAYESIAN PERSPECTIVE

TLDR
This paper suggests a class of at-tailed priors for a general likelihood function for robust Bayesian solutions, in the same spirit as the t-distribution being suggested as a at-tail prior for normal likelihood.

Approximate Models and Robust Decisions

Decisions based partly or solely on predictions from probabilistic models may be sensitive to model misspecification. Statisticians are taught from an early stage that "all models are wrong", but

Counterfactual Sensitivity and Robustness

TLDR
A tractable econometric framework for characterizing the sensitivity of counterfactuals with respect to a researcher's assumptions about the distribution of unobservables in a class of structural models is proposed and related to a measure of local sensitivity which is further characterized using an influence function representation.
...