Panel time series methods were born to address the issues of “long” panels of possibly nonstationary series, usually of macroeconomic nature. Such datasets, pooling together a sizable number of time series from different countries (regions, firms) have become increasingly common and are the main object of empirical research in many fields: development economics, regional or political science to name a few; the most typical unit of observation being a country or region within a reasonably large set of similar units and over at least two decades of either yearly or quarterly data.
Unlike “large” panels, the emphasis is therefore not only on ‐asymptotics but on both and tending to infinity, either sequentially or jointly (a seminal paper in this respect is Phillips and Moon, 1999). Specifying the order with which and diverge is essential for the properties of estimators.
The dynamics holds a more important, often prominent place (see e.g. Pesaran ...