Integration of macro and micro level forecasting methods for volatile environments: applications in durable goods industries

Abstract

This paper proposes an integrated forecasting system that combines an econometric method with a time series method and applies the system to the forecasting of Turkish monthly automobile sales data. An econometric model is used to generate initial base levels of quarterly sales forecasts upon which dynamic time series adjustments are made as actual monthly data are realised. Of particular concern are the effects of macro economic changes that occur more frequently in developing economies, thus, causing consumers to change their behaviour regarding major purchases. The proposed model considers long-term and short-term effects that influence sales of durable goods when economic disruptions are severe and/or temporary. The proposed integrative mechanism dynamically updates and revises medium-term econometric forecasts via short-term time series methods. Analysis of real life data supports the effectiveness of the proposed methodology in a growing-economy automotive market and simulation experiments facilitate generalisation of specific findings.