Forecasting the Relative Direction of Economic Growth by Using the Purchasing Managers` Index*

Abstract

This paper aims to predict the relative direction of the growth rate of Turkish real gross domestic product (GDP) by using the purchasing managers` index (PMI) data and to quantify the predictive performances of three alternative models: naive, constant and variable models. Naive model employs only the GDP data while the other two use both GDP and PMI data. The median growth rate is chosen as the main reference rate. In this case, the success rate of the forecasts by the variable (constant) model is 17.2 (3.4) percentage points higher than that of the naive model. The calculations for 8 other reference rates also lead to similar findings. The variable model has a higher success rate than the constant model at 4 out of 9 reference rates, while the constant model outpaces the variable model at only 2 reference rates. Moreover, the time-invariant PMI threshold that corresponds to a stagnant economic activity is found to be 47.5, any PMI reading above (below) which implies an expansion (a contraction)