Autoregressive process of order one Created 2026-09-28 Updated 2026-10-14
An autoregressive process of order one obeys
where is real and is centered white noise of variance . The zero-mean version is ; in the affine form , the stationary mean satisfies .
For , its unique weakly stationary process solution on all integer times is the causal time-series representation
The sum converges in the sense of mean-square convergence because . It is an infinite moving-average representation using only present and past driving noise. Weak white noise suffices for this second-order conclusion; strong white noise also gives strict stationarity. The autocovariance of an AR(1) process is , so its variance is and its ACF is . See Bartlett, lecture 5: AR(1) representation and causality.
The appeal of an AR(1) is that one coefficient expresses how a deviation persists. For , disturbances propagate with weights . For , these weights alternate sign. With , , and , the recursive forecasts of an autoregressive process give and one and two steps ahead. If , the stationary variance is , while the one-step forecast error variance is only . Under weak white noise these are best linear forecasts; independent driving noise also makes them conditional expectations given the observed past.
The condition describes a stationary causal solution, not every finite recursion with those coefficients. Starting from a prescribed value gives the transient in finite initialization of an AR(1) process. Nor is it a condition for all possible two-sided stationary solutions: when , noncausal stationary autoregression uses future driving noise. See stationary versus causal solution of a two-sided AR(1) equation for the boundary cases.
A causal time-series representation expresses a time series using present and past driving white noise. Its coefficients vanish at negative lags. Square summability ensures mean-square convergence for a white-noise input; absolute summability is the stronger stable-filter convention. Causality refers to the particular driving sequence, not merely to stationary existence.
Write for the driving white noise variance and . The autoregressive polynomial factors as
The discrete Gaussian white noise values are independent. The roots of are outside the unit disc, so expanding gives an absolutely summable causal time-series representation. Hence is a centered Gaussian process formed from driving white noise values at times at most . In particular, for , every driving white noise value with is independent of . The autocovariance tail recurrence of a causal ARMA process follows by taking covariance of the recursion with :
The roots of the characteristic polynomial of this tail recurrence are and . Let . There exist constants such that
For example, put and ; then and reproduce the two starting values and therefore the whole recurrence. Choose strictly larger than and all the finitely many values with . Symmetry now gives
For negative this is stronger than the requested bound with . It also proves .
For the Fourier calculation, take positive and positive. The supplied equal-norm trigonometric identities require ; the usual intended range is . Because is odd, at every such nonzero frequency. The geometric series of and over terms vanish. They imply
Other integer values of reduce modulo ; folding a negative frequency changes the sine coefficient only by a sign. The permitted zero frequency is handled below.
Grouping the double expectation sum according to the time lag yields
Replace the two truncated sums for lag by full sums. Their total is
Each replacement adds at most terms, each of absolute value at most one. This gives the finite-record Fourier covariance bound:
Both coefficients have expected value zero, so this expectation is their covariance.
For each finite , has a multivariate normal distribution because it is a linear transformation of the Gaussian observation vector. If the variances converge to , its characteristic function converges to
since the cross-covariance tends to zero. This is the characteristic function of two independent centered normal variables; zero limiting variances are allowed.
We can calculate the variance limits rather than merely assume their existence. Grouping the cosine covariance sum gives
Over a full record, . Truncation changes this by at most . The same statements hold for the sine products. Consequently
and the identical main term and bound apply to . Absolute summability now gives
Specify the spectral convention to identify this limit. Let
Here is the two-sided probability density function on , and is the one-sided probability density function on , satisfying . Thus each variance limit is . For this particular model, the filter representation also gives
For the specified squared-sum estimator, an exact expectation calculation uses :
The missing tail and the terms vanish; indeed the bias is by the exponential covariance bound. Therefore is asymptotically unbiased for the one-sided probability density function . Under the common two-sided convention its limit in expectation is , so is the appropriately normalized estimator.
At an interior frequency with , the joint Gaussian limit gives the exponential limit of a one-sided periodogram:
Moreover, for a centered Gaussian pair,
The nondegenerate exponential limiting law rules out convergence in probability to . Thus the unsmoothed periodogram is not consistent at frequencies with positive spectrum, despite its asymptotic unbiasedness. If , the nonnegative estimator has expectation tending to zero and is consistent there by Markov's inequality. Smoothing an increasing number of nearby ordinates while shrinking the frequency bandwidth can remove the positive-spectrum variance obstruction.
For , the sine coefficient is identically zero and the cosine coefficient is the normalized sum of the observations. The corrected limits are
The joint limit is , independent in the degenerate sense. The exact expectation formula for still applies and tends to , but now when , with limiting variance . The estimator is again inconsistent. This covers the zero frequency allowed by the printed inequality on .
Figure 1.
Normalized one-sided periodograms remain dispersed as the record length increases
. Details.
The figure uses the allowed cancellation , so the observations reduce to white noise. Orthogonal Fourier projections give the exponential law exactly at each plotted record length; increasing the record does not narrow the individual periodogram ordinate.
Define the autoregressive polynomial and the moving-average polynomial . The standard causality and invertibility root criteria for an ARMA model are
In the usual minimal autoregressive moving-average model, the two polynomials have no common roots. Under that convention, these are the necessary and sufficient root conditions for a causal and invertible representation with the given noise. Without minimality they remain sufficient; shared factors must be considered before claiming necessity.
To see why the conditions work, both ratios and are analytic on a disc of radius greater than one. Their power series coefficients therefore decay geometrically and are absolutely summable. The backshift operator then gives a well-defined mean-square convergent causal time-series representation
Filtering white noise by these fixed coefficients gives a weakly stationary process. The identity verifies the original equation. Multiplication by the absolutely summable inverse of also shows uniqueness among finite-variance stationary solutions. Similarly gives the invertible time-series representation.
Comparing coefficients in derives the first of the ARMA coefficient recursions:
Here for negative , , and for , so in particular . For the inverse filter, compare coefficients in . With , for , for , and for negative indices, this gives
The minus sign in the inverse recursion comes from moving the nonconstant moving-average terms to the other side.
For the first-order model, the standard uncancelled conditions are and . The geometric power series expansions give
Thus the ARMA(1,1) causal and inverse coefficients are
At these give and , including zero-valued parameters without any negative powers. If , the filtered process reduces to and all later coefficients vanish. This common factor cancellation in an ARMA model is why the root inequalities should not be advertised as necessary for every nonminimal parameterization. The stated inequalities still ensure uniqueness for the original equation as well as the desired causal and invertible filters.
A causal time-series representation uses only the present and past driving white noise. Thus the coefficient condition is
The series must have its stated convergence meaning. For centered white noise of positive finite variance, is sufficient and necessary for mean-square convergence. In the usual stable-filter convention one imposes the stronger . A bilateral stationary linear process need not be causal: terms with involve future driving values.