Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.
Ada: I'm Ada.
Emmy: And I'm Emmy, here with Kolya.
Ada: We're looking at the paper: "Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences".
Emmy: Here's the gist of the paper. Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences provides a finite-dimensional spectral characterisation for when the stationary covariance matrix of a finite-state Markov-switching affine recurrence exists.
Ada: This paper translates the mean-square spectral condition into an exact covariance-existence criterion for the stationary distribution.
Emmy: Pruning redundant reasoning tokens The primary focus of this research is to determine under what conditions the stationary solution of a strictly stationary Markov-switching affine recurrence is square-integrable, which ensures the existence of its variances, covariances, and Pearson correlations.
Ada: The authors address the fact that strict stationarity governed by a negative top Lyapunov exponent does not guarantee finite second-order objects.
Paper summary: Emmy: We’ve been looking at this paper "Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences" and its authors, Wilson Tai from Dublin City University. The thesis is quite specific: they provide a finite-dimensional spectral characterisation for when the stationary covariance matrix of a finite-state Markov-switching affine recurrence exists.
Ada: It claims that this existence hinges on the Markov-switching Kronecker operator T, specifically saying that under a natural Perron excitation condition on the innovation covariances, having the spectral radius of T less than one is both necessary and sufficient for square integrability.
Kolya: What does this mean in plain terms? It means we move past just checking if the process stays stationary along typical paths to checking a second-order spectral condition related to how those regimes interact over time.
Emmy: The authors show that this spectral check directly governs whether the regime-weighted second moments solve a finite-dimensional linear system, which lets you calculate that covariance matrix explicitly.
Ada: So, the implication is that covariance existence in these models depends on a second-order spectral check rather than just pathwise contraction alone. This distinction is crucial for anyone modeling these systems accurately.
Kolya: It’s interesting because they illustrate that strict stationarity and finite covariance are actually two different properties, and you can have one without the other if the system contracts along typical paths but has persistent visits to locally expansive regimes.
Emmy: That’s right, so for affine recurrences, the top Lyapunov exponent controls whether a stationary causal solution exists, but it's that Markov-switching Kronecker operator that controls whether that solution is square integrable.
Ada: The condition rho(T) less than one is the exact criterion for covariance existence when combined with Perron excitation, and they provide examples to show how persistent visits to expansive regimes can still destroy second moments.
Kolya: That’s a clear point on what's possible versus what's guaranteed by the standard contraction conditions. It sets a necessary check for any practical application of this theory where you need those variances and covariances.
Conclusion: Ada: So, we've been looking at this paper about Markov-switching affine recurrences and how to check if their covariance matrix even exists.
Emmy: Exactly, it’s all about that spectral characterisation of covariance existence in those models.
Ada: The authors are Wilson Tai from Dublin City University and they’ve given us a finite-dimensional way to check for that stationary covariance matrix.
Kolya: And the core idea is tying it all back to a specific operator, the Markov-switching Kronecker operator.
Emmy: It shows that we need this second-order spectral check, which is the spectral radius of that operator being less than one under certain conditions on innovation covariances.
Ada: So what does this actually mean for someone just listening? It means that just because a process is stationary doesn't guarantee you have finite variances and correlations.
Kolya: It highlights that you need to check this spectral condition in addition to the standard stationarity check when dealing with these switching models.
Emmy: The examples they run show that even if the system contracts along typical paths, persistent visits to certain regimes can still blow up those second-order objects.
Ada: That’s a crucial distinction because it shows that pathwise contraction isn't enough for covariance existence in this setup.
Kolya: It puts the onus on you to perform this spectral check—that condition being less than one—to ensure your covariance calculations actually yield finite numbers.
Emmy: It connects the top Lyapunov exponent, which deals with the long-term existence of a solution, with that operator that controls the second moments.
Ada: So it’s like having two separate checks in place for these Markov-switching affine models: one for existence, and one for having finite covariance.
Kolya: And if you're working on real data from a switching process, you absolutely need to be paying attention to that spectral radius check.
Emmy: It makes the theoretical link between stability and finite second moments very explicit in this framework.
Dublin City University DCU School of Mathematical Sciences
math.PR
Submitted: 2026-07-10
Updated: 2026-11-10
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 92/100
The gist: Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences provides a finite-dimensional spectral characterisation for when the stationary covariance matrix of a…
Key concepts
- Markov-switching Kronecker operator
- This is a block matrix used to describe how the regime-weighted second moments of the recurrence evolve over time. It captures the dynamics of these higher-order statistics, which are necessary to determine if variances and covariances remain finite.
- Spectral radius $ ho(T)$
- The spectral radius is a key number derived from this Kronecker operator. If this value is less than one, it ensures that the regime-weighted second moments do not grow unboundedly, which is the condition required for the covariance matrix to exist.
- Top Lyapunov exponent
- This exponent determines whether a strictly stationary causal solution exists for the recurrence. While its negativity guarantees stationarity (the process settles into a stable pattern), it does not guarantee that second-order moments like covariance are finite.
Terminology
Summary
Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences provides a finite-dimensional spectral characterisation for when the stationary covariance matrix of a finite-state Markov-switching affine recurrence exists. p. 0 This paper translates the mean-square spectral condition into an exact covariance-existence criterion for the stationary distribution. p. 0
Pruning redundant reasoning tokens
The primary focus of this research is to determine under what conditions the stationary solution of a strictly stationary Markov-switching affine recurrence is square-integrable, which ensures the existence of its variances, covariances, and Pearson correlations. The authors address the fact that strict stationarity governed by a negative top Lyapunov exponent does not guarantee finite second-order objects. p. 0 The key finding is that this existence is governed by a Markov-switching Kronecker operator, and the condition for square-integrability is related to the spectral radius of this operator. p. 3
The governing spectral objects
The evolution of regime-weighted second moments is described by a Markov-switching Kronecker operator, defined by the block matrix T where Tji:= pij (Aj ⊗ Aj). The spectral radius of this operator, denoted ρ(T), is central to the criterion for covariance existence. p. 0 Under a natural Perron-excitation condition on the innovation covariances, E∥Z∥2 < ∞ is necessary and sufficient for ρ(T) < 1. p. 0 This allows the paper to translate mean-square stability into an exact criterion for covariance existence in this context. p. 0
The necessity and sufficiency proof structure
The main result establishes equivalence between the square-integrability of the stationary solution and the spectral condition ρ(T) < 1. The sufficiency is proven by showing that if ρ(T) < 1, the second moments Mj are obtained from a finite-dimensional linear system, specifically Mj = X M i=1 pijAjMiA⊤ j + πjQj. p. 0 The necessity is shown by proving that if E∥Z∥2 < ∞, then the finite solution M = T M + q exists in the cone K, which implies ρ(T) < 1 via Lemma 3. p. 81.
Illustrating the separation of properties
The paper uses examples to demonstrate that strict stationarity and finite covariance are distinct properties. In a scalar example, the Lyapunov exponent can be negative while the second-order spectral radius exceeds one, showing that strict stationarity can persist even though variances and covariances are not finite objects. p. 0 A numerical illustration shows that a small increase in the locally expansive regime can push the second-order spectral radius above one while keeping the top Lyapunov exponent negative, leading to a stationary solution without finite covariance. p. 0
Conclusion and implications
The conclusion is that for affine recurrences, the top Lyapunov exponent controls the existence of a stationary causal solution, while the Markov-switching Kronecker operator controls whether that solution is square-integrable. Covariance calculations in Markov-switching affine models require this second-order spectral check in addition to the usual stationarity condition. p. 3 This distinction is crucial because it shows that covariance existence depends on a second-order spectral check rather than just pathwise contraction. p. 3 The paper also provides a scalar reduction showing EZ2 < ∞ ⇐⇒ ρ(P D2) < 1 under certain conditions. p. 0
References
[1] G. K. Basak and Z. p. 2-Q. Lu, Stationarity of switching VAR and other related models, arXiv preprint math/0507267, 2005. p. 2
[2] T. Bollerslev, Generalized autoregressive conditional heteroskedasticity, Journal of Econometrics 31 (1986), no. p. 5 3, 307–327.
[3] P. Bougerol and N. Picard, Strict stationarity of generalized autoregressive processes, The Annals of Probability 20 (1992), no. p. 0 4, 1714–1730.
[4] P. Bougerol and N. Picard, Stationarity of GARCH processes and of some nonnegative time series, Journal of Econometrics 52 (1992), no. p. 3 1–2, 115–127.
[5] A. Brandt, The stochastic equation Yn+1 = AnYn +Bn with stationary coefficients, Advances in Applied Probability 18 (1986), no. p. 4 1, 211–220.
[6] D. Buraczewski, E. Damek and T. Mikosch, Stochastic Models with Power-Law Tails: The Equation X = AX + B, Springer, New York, 2016. p. 3
[7] O. L. V. Costa, M. D. Fragoso and R. P. Marques, Discrete-Time Markov Jump Linear Systems, Springer, London, 2005. p. 3
[8] P. Diaconis and D. Freedman, Iterated random functions, SIAM Review 41 (1999), no. p. 2 1, 45–76.
[9] R. F. Engle, Autoregressive conditional heteroscedasticity with estimates of the variance of United Kingdom inflation, Econometrica 50 (1982), no. 4, 987–1007.
[10] C. M. Goldie, Implicit renewal theory and tails of solutions of random equations, The Annals of Applied Probability 1 (1991), no. p. 3 1, 126–166.
[13] H. Kesten, Random difference equations and renewal theory for products of random matrices, Acta Mathematica 131 (1973), 207–248. p. 2
[14] H.-M. Krolzig, Markov-Switching Vector Autoregressions: Modelling, Statistical Inference, and Application to Business Cycle Analysis, Lecture Notes in Economics and Mathematical Systems, vol. p. 2 454, Springer, Berlin, 1997.
[15] S. Ling, L. Peng and F. Zhu, Inference for a special bilinear time series model, Journal of Time Series Analysis 36 (2015), no. p. 2 1, 61–76.
[16] X. Mao and C. Yuan, Stochastic Differential Equations with Markovian Switching, Imperial College Press, London, 2006. p. 1
[17] S. P. Meyn and R. L. Tweedie, Markov Chains and Stochastic Stability, 2nd ed. p. 1, Cambridge University Press, Cambridge, 2009. p. 0
[18] D. B. Nelson, Stationarity and persistence in the GARCH(1,1) model, Econometric Theory 6 (1990), no. p. 3 3, 318–334.
[19] D. F. Nicholls and B. G. Quinn, Random Coefficient Autoregressive Models: An Introduction, Springer, New York, 1982. p. 3
[20] J. Shao and F. Xi, Stability and recurrence of regime-switching diffusion processes, SIAM Journal on Control and Optimization 52 (2014), no. p. 1 6, 3496–3516.
[21] T. Subba Rao and M. M. Gabr, An Introduction to Bispectral Analysis and Bilinear Time Series Models, Lecture Notes in Statistics, vol. p. 2 24, Springer, New York, 1984. p. 5
[22] W. Vervaat, On a stochastic difference equation and a representation of non-negative infinitely divisible random variables, Advances in Applied Probability 11 (1979), no. p. 2 4, 750–783.
[23] G. G. Yin and C. Zhu, Hybrid Switching Diffusions: Properties and Applications, Springer, New York, 2010. p. 2
--- Page 1 ---
Spectral Characterisation of Covariance Existence in Markov-Switching Affine Recurrences Wilson Tai Dublin City University DCU School of Mathematical Sciences Abstract This paper provides a finite-dimensional spectral characterisation for when the stationary covariance matrix of a finite-state Markov-switching affine recurrence exists.
--- Page 3 ---
1 Introduction A strictly stationary Markov-switching affine recurrence need not have a finite stationary covariance matrix. It is therefore natural to ask under what additional conditions these second-order objects exist. p. 2 The relevant object is the Markov-switching Kronecker operator. p. 2
How the training signal is built
The evolution of regime-weighted second moments is governed by a Markov-switching Kronecker operator. This operator describes the evolution of regime-weighted second moments. p. 2 p. 2 The operator is represented by the block matrix T with blocks Tji = pij (Aj ⊗ Aj), 1 ≤ i, j ≤ M. p. 6 This operator describes the evolution of regime-weighted second moments.
Where the gains come from
The main result is that, under a natural Perron-excitation condition on the innovation covariances, ρ(T) < 1 is necessary and sufficient for the stationary solution to be square-integrable. When this condition holds, the regime-weighted second moments solve a finite-dimensional linear system, which yields the covariance matrix explicitly. p. 0
The final criterion
The top Lyapunov exponent controls the existence of a stationary causal solution, and the Markov-switching Kronecker operator controls whether that stationary solution is square-integrable. The condition ρ(T) < 1 is the exact criterion for covariance existence when combined with Perron excitation. p. 3 p. 3 The examples illustrate that a recurrence may contract along typical sample paths and hence admit a strictly stationary solution, while sufficiently persistent visits to locally expansive regimes can still destroy second moments. p. 2 p. 2
Conclusion
The top Lyapunov exponent controls the existence of a stationary causal solution, and the Markov-switching Kronecker operator controls whether that stationary solution is square-integrable. The condition ρ(T) < 1 is the exact criterion for covariance existence when combined with Perron excitation. The examples illustrate that a recurrence may contract along typical sample paths and hence admit a strictly stationary solution, while sufficiently persistent visits to locally expansive regimes can still destroy second moments. The point is that covariance calculations in Markov-switching affine models require a second-order spectral check in addition to the usual stationarity condition. p. 3
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved AI system can determine if a stationary solution of a Markov-switching affine recurrence has finite variances, covariances, and Pearson correlations by calculating the spectral radius of the Markov-switching Kronecker operator T.
-
The improved system can explicitly reconstruct the covariance matrix Sigma if the spectral radius condition rho(T) less than 1 holds, using the formula Mj = X M i=1 pij Aj MiA top j + pi j Q j for each regime j.
-
The system can distinguish between strict stationarity and finite covariance by comparing the top Lyapunov exponent (governing strict stationarity) with the second-order spectral radius rho(T) (governing covariance existence).
Sources
Related papers
- Rank-One HCIZ Integrals and Multiplicative Convolutions
- Repeated averaging on random d-regular graphs
- Precise cover times for branching random walks on Hamming graphs: (iterated) logarithmic corrections
- Markov Chain CLTs: Resolving Open Problems
- Marked stable L'evy processes in Brownian local time
- A Sharp Refinement of Samuels' Inequality via Cap Moves