Inferring causal impact using Bayesian structural time-series models

DOI10.1214/14-AOAS788zbMATH Open1454.62473arXiv1506.00356OpenAlexW2078639378MaRDI QIDQ89980FDOQ89980

Fabian Gallusser, Fabian Gallusser, Nicolas Remy, Steven L. Scott, Steven L. Scott, Kay H. Brodersen, Jim Koehler, Nicolas Remy, Kay H. Brodersen, Jim Koehler

Publication date: 1 March 2015

Published in: The Annals of Applied Statistics (Search for Journal in Brave)

Abstract: An important problem in econometrics and marketing is to infer the causal impact that a designed market intervention has exerted on an outcome metric over time. This paper proposes to infer causal impact on the basis of a diffusion-regression state-space model that predicts the counterfactual market response in a synthetic control that would have occurred had no intervention taken place. In contrast to classical difference-in-differences schemes, state-space models make it possible to (i) infer the temporal evolution of attributable impact, (ii) incorporate empirical priors on the parameters in a fully Bayesian treatment, and (iii) flexibly accommodate multiple sources of variation, including local trends, seasonality and the time-varying influence of contemporaneous covariates. Using a Markov chain Monte Carlo algorithm for posterior inference, we illustrate the statistical properties of our approach on simulated data. We then demonstrate its practical utility by estimating the causal effect of an online advertising campaign on search-related site visits. We discuss the strengths and limitations of state-space models in enabling causal attribution in those settings where a randomised experiment is unavailable. The CausalImpact R package provides an implementation of our approach.

Full work available at URL: https://arxiv.org/abs/1506.00356

zbMATH Keywords

causal inferenceadvertisingcounterfactualdifference in differenceseconometricsmarket researchobservationalsynthetic control

Mathematics Subject Classification ID

Bayesian inference (62F15) Applications of statistics to economics (62P20) Inference from stochastic processes and prediction (62M20)

Cites Work

• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Predicting the present with Bayesian structural time series
• DATA AUGMENTATION AND DYNAMIC LINEAR MODELS
• On Gibbs sampling for state space models
• A simple and efficient simulation smoother for state space time series analysis
• Determining the optimal return on investment for an advertising compaign
• Estimating the causal effects of marketing interventions using propensity score methodology
• Data augmentation for support vector machines
• Asymptotic properties of a robust variance matrix estimator for panel data when \(T\) is large
• Semiparametric Difference-in-Differences Estimators
• Rao–Blackwellization for Bayesian Variable Selection and Model Averaging in Linear and Binary Regression: A Novel Data Augmentation Approach
• Mixtures of g Priors for Bayesian Variable Selection
• How Much Should We Trust Differences-In-Differences Estimates?
• Generalized least squares inference in panel and multilevel models with serial correlation and fixed effects
• The simulation smoother for time series models
• Identification and Inference in Nonlinear Difference-in-Differences Models
• Bayes and empirical-Bayes multiplicity adjustment in the variable-selection problem
• Bayesian forecasting and dynamic models.

Cited In (29)

• Extending Bayesian structural time-series estimates of causal impact to many-household conservation initiatives
• A machine learning model for nowcasting epidemic incidence
• Assessing the effect of advertising expenditures upon sales: a Bayesian structural time series model
• Discussion of ‘Dynamic dependence networks: financial time series forecasting and portfolio decisions’
• Invariance, causality and robustness
• Bayesian matrix completion approach to causal inference with panel data
• Counterfactual Analysis With Artificial Controls: Inference, High Dimensions, and Nonstationarity
• Robust causal inference for incremental return on ad spend with randomized paired geo experiments
• Bayesian method for causal inference in spatially-correlated multivariate time series
• Synthetic learner: model-free inference on treatments over time
• Synthetic Control with Time Varying Coefficients A State Space Approach with Bayesian Shrinkage
• Does terrorism trigger online hate speech? On the association of events and time series
• CausalImpact
• Confidence intervals of treatment effects in panel data models with interactive fixed effects
• Posterior predictive treatment assignment methods for causal inference in the context of time-varying treatments
• Analysis of telecom service operation behavior with time series
• Title not available (Why is that?)
• Assessing the causal effect of binary interventions from observational panel data with few treated units
• Statistical modeling of causal effects in continuous time
• Estimating the effects of a California gun control program with multitask Gaussian processes
• Estimating the effectiveness of permanent price reductions for competing products using multivariate Bayesian structural time series models
• Time Series Experiments and Causal Estimands: Exact Randomization Tests and Trading
• SPORTSCausal
• Title not available (Why is that?)
• Title not available (Why is that?)
• A dynamic causal modeling of the second outbreak of COVID-19 in Italy
• Demand forecasting of individual probability density functions with machine learning
• Title not available (Why is that?)
• Statistical modeling of computer malware propagation dynamics in cyberspace

Uses Software

• R
• CausalImpact

This page was built for publication: Inferring causal impact using Bayesian structural time-series models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q89980)