RNN‐based counterfactual prediction, with an application to homestead policy and public schooling (2017-12-10T00:00:00.000000Z)

TL;DR

The method for estimating the long‐run impact of US homestead policy on public school spending and the learned representation of control units is applied to the treated units for predicting counterfactual outcomes.

Abstract

This paper proposes a method for estimating the effect of a policy intervention on an outcome over time. We train recurrent neural networks (RNNs) on the history of control unit outcomes to learn a useful representation for predicting future outcomes. The learned representation of control units is then applied to the treated units for predicting counterfactual outcomes. RNNs are specifically structured to exploit temporal dependencies in panel data and are able to learn negative and non‐linear interactions between control unit outcomes. We apply the method to the problem of estimating the long‐run impact of US homestead policy on public school spending.

Authors

Jason Poulos

2 papers

Shuxi Zeng

1 papers

TL;DR

Abstract

Authors

References82 items

Synthetic controls with staggered adoption

Deep Learning for Natural Language Processing

A Bayesian Alternative to Synthetic Control for Comparative Case Studies

Nation-building Through Compulsory Schooling during the Age of Mass Migration

Deep Generalized Method of Moments for Instrumental Variable Analysis

Synthetic Difference in Differences

The Augmented Synthetic Control Method

Synthetic Control Method: Inference, Sensitivity Analysis and Confidence Sets

Design-Based Analysis in Difference-in-Differences Settings with Staggered Adoption

Synthetic Control and Inference

Robust Synthetic Control

Matrix Completion Methods for Causal Panel Data Models

R2N2: Residual Recurrent Neural Networks for Multivariate Time Series Forecasting

Deep and Confident Prediction for Time Series at Uber

Deep IV: A Flexible Approach for Counterfactual Prediction

Double/Debiased Machine Learning for Treatment and Structural Parameters

Position-Based Content Attention for Time Series Forecasting with Sequence-to-Sequence RNNs

Revisiting the synthetic control estimator

Imputation with the R Package VIM

Balancing, Regression, Difference-in-Differences and Synthetic Control Methods: A Synthesis

Generalized Synthetic Control Method: Causal Inference with Interactive Fixed Effects Models

ARCO: An Artificial Counterfactual Approach for High-Dimensional Panel Time-Series Data

Approximate residual balancing: debiased inference of average treatment effects in high dimensions

A Theoretically Grounded Application of Dropout in Recurrent Neural Networks

Attention-Based Models for Speech Recognition

Pooling Multiple Case Studies Using Synthetic Controls: An Application to Minimum Wage Policies

Inferring causal impact using Bayesian structural time-series models

Grammar as a Foreign Language

Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling

Neural Machine Translation by Jointly Learning to Align and Translate

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

Balancing Covariates via Propensity Score Weighting

Comparative Politics and the Synthetic Control Method

A Blockwise Descent Algorithm for Group-penalized Multiresponse and Multinomial Regression

Program evaluation and causal inference with high-dimensional data

The Political Economy of the American Frontier

On the Use of Geographic Information Systems in Economic History: The American Transportation Revolution Revisited

Nation-Building and Education

On the difficulty of training recurrent neural networks

Oracle Inequalities for High Dimensional Vector Autoregressions

Supervised Sequence Labelling with Recurrent Neural Networks

Oracle Efficient Estimation and Forecasting with the Adaptive Lasso and the Adaptive Group Lasso in Vector Autoregressions

MICE: Multivariate Imputation by Chained Equations in R

MissForest - non-parametric missing value imputation for mixed-type data

Causal inference

Strong rules for discarding predictors in lasso‐type problems

Synthetic Control Methods for Comparative Case Studies: Estimating the Effect of California’s Tobacco Control Program

Understanding the difficulty of training deep feedforward neural networks

Regularization Paths for Generalized Linear Models via Coordinate Descent.

Inequality in Land Ownership, the Emergence of Human Capital Promoting Institutions and the Great Divergence

Persistence of Power, Elites, and Institutions

Doubly Robust Estimation in Missing Data and Causal Inference Models

The Economic Costs of Conflict: A Case Study of the

The Evolution of Suffrage Institutions in the New World

How Much Should We Trust Differences-in-Differences Estimates?

Digest of Education Statistics 2002

The Mediating Effect of Npd-Activities and Npd-Performance on the Relationship between Market Orientation and Organizational Performance

Exponentiated Gradient Versus Gradient Descent for Linear Predictors

Neural networks

Hierarchical Recurrent Neural Networks for Long-Term Dependencies

Digest of Education Statistics

[On the Application of Probability Theory to Agricultural Experiments. Essay on Principles. Section 9.] Comment: Neyman (1923) and Causal Inference in Experiments and Observational Studies

Public Education as Nation-Building in America: Enrollments and Bureaucratization in the American States, 1870-1930

Estimating causal effects of treatments in randomized and nonrandomized studies.

A General Coefficient of Similarity and Some of Its Properties

DEEP NEURAL NETWORKS FOR ESTIMATION AND INFERENCE

General Land Oﬃce

imputeTS: Time Series Missing Value Imputation in R

General Land Office (GLO) Records Automation

Deep Learning

Random forests iterative imputation method based on random forests; implemented with the missForest R package

Data base on historical ﬁnances of federal, state and local governments

The origins of state capacity: Property rights, taxation and politics

Comment : Neyman ( 1923 ) and Causal Inference in Experiments and Observational Studies

NATIONAL CENTER FOR EDUCATION STATISTICS

Regression Shrinkage and Selection via the Lasso