Do We Really Need Filters In Estimating Output Gap? : Evidence From Turkey

We estimate an output gap indicator for Turkey without resorting to any kind of a filtering procedure. Our approach stands on a two-step procedure : First, we pick such variables that are directly informative about the phase of the business cycle, where the decision of choice depends on their statistical and economic significance in estimated Phillips curves. Second, we model business cycles as the common driver of the selected variables and estimate it in a small scale dynamic factor model setting. In this way, we produce a filter-free measure of output gap, which proves to be superior to any other filter-based measure as being immune to end-sample revisions. Using up-to-date survey-based variables instead of filtered macroeconomic aggregates, we not only postulate a way of avoiding revision uncertainty embodied in statistical filters, but also meet the need for timely information as we deliver information on the cyclical position of the economy two-quarters in advance of the GDP.