SpinSVAR: Estimating Structural Vector Autoregression Assuming Sparse Input
This addresses the challenge of SVAR estimation for time-series analysis, particularly in financial applications like stock market data, with incremental improvements through sparse input modeling.
The paper tackles the problem of estimating structural vector autoregression (SVAR) from time-series data by assuming sparse inputs, introducing SpinSVAR as a novel method that models inputs as independent Laplacian variables for sparsity and uses maximum likelihood estimation based on least absolute error regression. The result shows that SpinSVAR outperforms state-of-the-art methods in accuracy and runtime on synthetic data and successfully clusters stocks and identifies structural shocks in S&P 500 data.
We introduce SpinSVAR, a novel method for estimating a structural vector autoregression (SVAR) from time-series data under sparse input assumption. Unlike prior approaches using Gaussian noise, we model the input as independent Laplacian variables, enforcing sparsity and yielding a maximum likelihood estimator (MLE) based on least absolute error regression. We provide theoretical consistency guarantees for the MLE under mild assumptions. SpinSVAR is efficient: it can leverage GPU acceleration to scale to thousands of nodes. On synthetic data with Laplacian or Bernoulli-uniform inputs, SpinSVAR outperforms state-of-the-art methods in accuracy and runtime. When applied to S&P 500 data, it clusters stocks by sectors and identifies significant structural shocks linked to major price movements, demonstrating the viability of our sparse input assumption.