A COMPOSITE LEADING INDICATOR CALCULATION FOR TURKISH EXPORTS

Leading indicators are series that are examined to forecast fluctuations in a reference series. The series which are selected as the leading indicator of reference series are combined using weighted averages to form “composite leading indicators”. In this study, we aim to form a “composite leading indicator” to forecast Turkey’s exports series. For this purpose candidate leading indicator series were seasonally adjusted by using “TRAMO/SEATS” and de-trended by using Hodrik-Prescott filter. Best performing series were selected by looking at noise/signal ratios, improvement ratios, leading and coincident correlations and Granger causality relationships. Afterwards, a weighted average of best performing series were taken and thus, “a compostie leading indicator for Turkey’s exports” was formed. The weights used are noise/signal ratio and improvement ratio. Calculated composite leading indicator improves the forecast of important declines in Turkey’s exports up to 68%.