The claim
Maximising the General Predictive Likelihood yields parameter estimates identical to those of the conventional one-step-ahead estimator MSE_1 under correct model specification.
Scope: conditions: correctly specified linear state space models with Gaussian errors; not for: misspecified models where |\hat{A}| \neq 1 · Proof: E2 · Uses: C7, C13 · Assumes: C19, C21 · Limitations: L3 · Extends: B5, p. 77 · Verification: not checked
Claims it rests on
In an SSOE state space model with independent homoscedastic errors, the covariance between -step-ahead and -step-ahead forecast errors is given by so multi-step forecast errors have non-zero cross-horizon covariance whenever .
Proof: E1 · Assumes: C19 · Contradicts: B8, p. 11 · Verification: not checked · Used by: C3
The logarithm of the Generalised Variance decomposes as , where under correct model specification.
Proof: E2 · Assumes: C19, C21 · Verification: not checked · Used by: C3
Evidence
Analytical derivation in the single source of error (SSOE) state space framework showing that multi-step loss functions minimise multi-step forecast error variances and covariances, shrinking dynamic state-space parameters toward zero (and regression parameters not at all because ), along with specific derivations for ARIMA(1,1,1) and pure additive ETS models (Sections 2, 3, and Appendices A–B).
Foundation: Kolmogorov probability axioms and linear algebra
- \eta_{t+h|t} = y_{t+h} - \mu_{t+h|t} = \sum_{j=1}^{h-1} c_j \epsilon_{t+j} + \epsilon_{t+h}, \quad c_j = w' F^{j-1} g (definition; from D1)
- \sigma_h^2 = \sigma_1^2 \left(1 + \sum_{j=1}^{h-1} c_j^2\right), \quad \sigma_{i,j} = \sigma_1^2 \left(c_{|i-j|} + \sum_{l=1}^{i-1} c_{|l-j|} c_{|i-l|}\right) \text{ for } i,j > 1 (variance and covariance of linear combinations; from S1, C19)
- \text{Minimising } MSE_h, TMSE, \text{ or } MSCE \text{ minimises } s_1^2 \text{ and } \hat{c}_j^2 \text{ toward } 0 \text{ for } h > 1, \text{ whereas } g=0 \text{ in static regression implies } c_j = 0 \text{ and no shrinkage.} (sample substitution and minimisation; from S2)
Analytical derivation of concentrated predictive log-likelihoods for MSE_h, MSCE, and the General Predictive Likelihood (GPL), proving that log GV = log s_1^2 + log |\hat{A}| and that |\hat{A}| = 1 under correct specification, and defining GTMSE via diagonal restriction on \hat{\Sigma} (Section 4 and Appendices C–E).
Foundation: Kolmogorov probability axioms and multivariate normal likelihood theory
- l_{GPL_h}(\theta, \hat{\Sigma} \mid Y) = -\frac{T-h}{2}\left(h \log(2\pi e) + \log|\hat{\Sigma}|\right), \quad GV = |\hat{\Sigma}| (concentrating multivariate normal log-likelihood; from D7, C21)
- \log GV = \log s_1^2 + \log|\hat{A}|, \quad |\hat{A}| = 1 \text{ under correct model specification} (matrix determinant reduction; from S1)
- GTMSE \propto \sum_{j=1}^{h} \log(s_j^2) = h \log(s_1^2) + \sum_{j=2}^{h} \log\left(1 + \sum_{i=1}^{j-1} \hat{c}_i^2\right) (diagonal restriction and logarithmic scaling; from S1, D6)
Limitations
The equivalence between General Predictive Likelihood and one-step-ahead MSE_1 relies on correct model specification so that ; under model misspecification, off-diagonal covariances do not cancel and GPL estimates diverge from MSE_1.
Assumptions
Definitions
Methods 4
D2 MSE_1: Conventional one-step-ahead mean squared error estimator minimising where is the one-step-ahead in-sample forecast error over observations.
D6 GTMSE: Geometric Trace Mean Squared Error estimator introduced in this paper, minimising the sum of logarithmic multi-step mean squared errors across horizons to :
D7 General Predictive Likelihood: General Predictive Likelihood for consecutive observations under Gaussian errors, whose concentrated log-likelihood is where is the sample multi-step forecast error covariance matrix.
D8 Generalised Variance: Determinant of the sample multi-step forecast error covariance matrix, whose minimisation is equivalent to maximising the General Predictive Likelihood.
Theories 1
D1 SSOE state space model: An additive-error linear state space model of the form where is the observed value at time , is the state vector, is the transition matrix, is the persistence vector, is the measurement vector, and , with multi-step coefficients . Source: Snyder (1985).
Context
References
- Snyder, R. D., 1985. Recursive Estimation of Dynamic Linear Models. Journal of the Royal Statistical Society Series B: Statistical Methodology 47, 272–276. https://doi.org/10.1111/j.2517-6161.1985.tb01355.x Used by: D1
Score
Scores are computed by rules from the claimset file (EntryLib rubric 0.7): they check the record, not the science, and may change as the rubric and reviews develop. The rubric.
C3: 61, Partly supported
Novelty: new, by the authors
- −25 Evidence Rests on an unchecked proof; evidence named without the observations the claim reports. To remove it: Support the claim with evidence the claimset records, citing the result entries or table rows it reports, or have its proof checked; for cited work, read the work in full and quote the passage the claim rests on from a captured copy (claimset extension review).
- −10 Open problems Limitation L3 (could reverse, not mitigated): The equivalence between General Predictive Likelihood and one-step-ahead MSE_1 relies on correct model specification so that ; under model misspecification, off-diagonal covariances do not cancel and GPL estimates diverge from MSE_1. To remove it: Mitigate it with evidence, or narrow the claim's scope so that it no longer applies.
- −3 Open problems Assumption C19 is not checked: The one-step-ahead innovation errors of the state space model are independent and identically distributed with zero mean and constant variance . To remove it: Test the assumption, or analyse how sensitive the claim is to it.
- −1 Open problems Assumption C21 is not checked: The one-step-ahead innovation errors follow a Normal distribution . To remove it: Test the assumption, or analyse how sensitive the claim is to it.
C7: 72, Partly supported
Novelty: new, by the authors
- −25 Evidence Rests on an unchecked proof; evidence named without the observations the claim reports. To remove it: Support the claim with evidence the claimset records, citing the result entries or table rows it reports, or have its proof checked; for cited work, read the work in full and quote the passage the claim rests on from a captured copy (claimset extension review).
- −3 Open problems Assumption C19 is not checked: The one-step-ahead innovation errors of the state space model are independent and identically distributed with zero mean and constant variance . To remove it: Test the assumption, or analyse how sensitive the claim is to it.
C13: 71, Partly supported
Novelty: new, by the authors
- −25 Evidence Rests on an unchecked proof; evidence named without the observations the claim reports. To remove it: Support the claim with evidence the claimset records, citing the result entries or table rows it reports, or have its proof checked; for cited work, read the work in full and quote the passage the claim rests on from a captured copy (claimset extension review).
- −3 Open problems Assumption C19 is not checked: The one-step-ahead innovation errors of the state space model are independent and identically distributed with zero mean and constant variance . To remove it: Test the assumption, or analyse how sensitive the claim is to it.
- −1 Open problems Assumption C21 is not checked: The one-step-ahead innovation errors follow a Normal distribution . To remove it: Test the assumption, or analyse how sensitive the claim is to it.