Paper: arXiv 2312.17061
Authors: Parley R Yang, Alexander Y Shestopaloff
Abstract
Vector Error Correction Model (VECM) is a classic method to analyse cointegration relationships amongst multivariate non-stationary time series. In this paper, we focus on high dimensional setting and seek for sample-size-efficient methodology to determine the level of cointegration. Our investigation centres at a Bayesian approach to analyse the cointegration matrix, henceforth determining the cointegration rank. We design two algorithms and implement them on simulated examples, yielding promising results particularly when dealing with high number of variables and relatively low number of observations. Furthermore, we extend this methodology to empirically investigate the constituents of the S&P 500 index, where low-volatility portfolios can be found during both in-sample training and out-of-sample testing periods.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs not modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 6.5)
Why this score: The paper presents advanced Bayesian theory with complex priors, asymptotic analysis, and multi-step derivations for high-dimensional time series, earning a high math score. It includes empirical validation on S&P 500 data with in-sample and out-of-sample portfolio results, but lacks fully detailed backtesting code or implementation specifics, resulting in moderate empirical rigor.
Research Flowchart
flowchart TD A["Research Goal<br>Determine cointegration rank in<br>high-dim VECM"] --> B["Data Inputs<br>Simulated & S&P 500 Data"] B --> C["Key Methodology<br>Bayesian Inference on<br>Cointegration Matrix"] C --> D["Computational Process<br>Two Designed Algorithms<br>(High dim / Low sample)"] D --> E["Key Findings<br>Accurate rank detection &<br>Low-volatility portfolios"]
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a Bayesian approach with two algorithms for high‑dimensional VECM cointegration rank estimation.
Source sentence
We design two algorithms and implement them on simulated examples, yielding promising results particularly when dealing with high number of variables and relatively low number of observations.
Assumes 5
- All series are I(1) non‑stationary. model
Source sentence
When Y (·, j) is non‑stationary but when ∆Y (·, j) defined by ∆Y (t, j) := Y (t, j) −Y (t −1, j) is stationary, we write Y (·, j) ∼I(1).
- Feasibility condition T > pr for estimating non‑sparse coordinates of Π. statistical
Source sentence
it is reasonable to impose a feasibility assumption where T > pr.
- Gaussian error distribution for the data. statistical
Source sentence
The first two terms of Equation 9 are due to Gaussian assumption (Equation 7) with Jeffery prior on the variance.
- Sparse structure of the cointegration matrix via SSL prior. model
Source sentence
The penalty term pen(R, θ) is due to the SSL assumption (Equation 8).
Show 1 more
- Low‑volatility portfolios are stable over training and testing periods. market
Source sentence
Our algorithms yield various independent and insightful portfolios that yield low volatilities, as co‑integration theory would presume.
Data
- Synthetic VECM simulations (p=10, T=100, r∈[5]) synthetic public
Source sentence
In the first regime, we consider p = 10, T = 100, and we scan the cases for r ∈[5].
- Synthetic VECM simulations (p=30, T=300, r∈[5]) synthetic public
Source sentence
In the second regime, we consider p = 30, T = 300, and we scan the cases for r ∈[5].
Costs
Not modeled
Finds 4
- Algorithm 1 achieves 100% correct rank detection when r=0 and 65% correct when r=5 for p=30. performance
Source sentence
... when r = 5, the algorithm detects only 65% correct in terms of ˆr = 5 for p = 30
- Algorithm 2 correctly estimates rank in all simulated scenarios when the rank range is relaxed. performance
Source sentence
The algorithm 2 further estimate all scenarios correctly when we relax the range into between r −p/50 and r + p/50.
- Algorithm 2 outperforms Algorithm 1 by appropriately halting λ0 increments and providing probabilistic rank inference. theoretical
Source sentence
significantly outperforms algorithm 1 by halting the λ0 increment appropriately.
- Empirical application yields low‑volatility, stable portfolios based on estimated cointegration vectors. performance
Source sentence
Our algorithms yield various independent and insightful portfolios that yield low volatilities, as co‑integration theory would presume.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 15 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (11)
- Large Spillover Networks of Nonstationary Systems (2022) — link
- Spike-and-Slab Meets LASSO: A Review of the Spike-and-Slab LASSO (2020) — link
- Determination of vector error correction models in high dimensions (2019) — link
- Bayesian estimation of sparse signals with a continuous spike-and-slab prior (2018) — link
- The Spike-and-Slab LASSO (2018) — link
- Needles and Straw in a Haystack: Posterior concentration for possibly sparse sequences (2012) — link
- Bayesian Time Series Models (2011) — link
- Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models (1991) — link
and 3 more
0 of 11 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.