The current time series paradigm - with its obsession with logocentric representations of unit roots, a strangely paradoxical notion in the sense that its 'otherness' with its I(0) counterpart depends usually on a simple first differencing operation - finds itself unable to deal with the essentially Heideggerian notion of the Dasein, wherein like the false subject/object distinction, the stationarity/nonstationarity bipolar opposition finds itself irrelevant in light of the comprehensive post-structural approach to time series through unobserved component modelling, wherein non-stationarity can be dealt with by appropriate post-modern filters, unlike the antiquated Wiener filter with its obvious and abhorrent phallocentrism and restriction to stationary series unlike the feminine and outcast non-stationary series which are seen - through their aphallism - as not worthy of contemplation. Of course, many scholars sort to bridge the divide through cointegration - that half-way house between what was seen as "spurious" regression (ascribed in light of the supposedly hysterical, female qualities of non-stationarity) and masculine first-differencing. Such commonality can of course be accommodated by post-structural time-series through appropriate restrictions on the deconstructive-covariance matrix. In a broader multivariate, almost post-religious Nietzschian sense, I would hasten to add that there is nothing outside the covariance matrix.
- Jacques Derrida, the lost papers