Regularized Estimation and Testing for High-Dimensional Multi-Block Vector-Autoregressive Models
Authors: Jiahe Lin, George Michailidis
JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The performance of the model and the testing procedures are evaluated on synthetic data, and illustrated on a data set involving log-returns of the US S&P100 component stocks and key macroeconomic variables for the 2001 16 period. Section 5 contains selected numerical results that assess the performance of the estimation and testing procedures. Finally, an application to financial and macroeconomic data that was previously discussed as motivation for the model under consideration is presented in Section 6. |
| Researcher Affiliation | Academia | Jiahe Lin EMAIL Department of Statistics University of Michigan Ann Arbor, MI 48109, USA George Michailidis EMAIL Department of Statistics and the Informatics Institute University of Florida Gainesville, FL 32611, USA |
| Pseudocode | Yes | Algorithm 1: Computational procedure for estimating A and Σ 1 u . Algorithm 2: Computational procedure for estimating B, C and Σ 1 v . |
| Open Source Code | No | The text does not explicitly provide a link to a source code repository or state that the code for the methodology is openly available. The provided URL points to the paper's webpage, which may or may not contain code. |
| Open Datasets | No | The paper states that 'The complete lists of stocks and macroeconomic variables, together with the preprocessing to ensure stationarity used in this study are given in Supplementary Material ??'. However, it does not provide a direct link, DOI, repository, or explicit statement of public availability for the dataset files used in the experiments. |
| Dataset Splits | Yes | Analogously to the strategy employed by Billio et al. (2012), we consider 36-month-long rolling-windows for fitting the model Xt = AXt 1 + Ut, for a total of 143 estimates of the transition matrix A. Based on the previous findings, we partition the time frame spanning 2001-2016 into the following periods: pre(2001/07 2007/03), during(2007/01 2009/12) and post-crisis (2010/01-2016/06) one. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., GPU/CPU models, memory) used to conduct the experiments or simulations. |
| Software Dependencies | No | The paper mentions methods like 'graphical Lasso' and 'Lasso penalty' but does not specify any particular software libraries, programming languages, or their version numbers used for implementation. |
| Experiment Setup | Yes | Tuning parameters are chosen based on BIC. For the sparse components, each entry in A and C is nonzero with probability 2/p1 and 1/p2 respectively, and the nonzero entries are generated from Unif ([ 2.5, 1.5] [1.5, 2.5]), then scaled down so that the spectral radii ρ(A) and ρ(C) satisfy the stability condition. tuning parameters based on a search over a 10 x 10 lattice (with (λB, λC) [0.5, 4] [0.2, 2], equal-spaced) using the BIC. stability selection (Meinshausen and B uhlmann, 2010), with the threshold set at 0.6 for including an edge in A. |