Forecasting Inflation: A Combination Approach


  • Martin Bojaj University of Montenegro, Faculty of Economics
  • Gordana Djurovic University of Montenegro



Macroeconometric Forecasting; Inflation; Forecast Combination; Impulse Response Functions; Inverse MSE Approach.


The objective of this paper is to investigate and forecast the determinants of Montenegrin inflation empirically, using forecast combination methods, from January 2006 to December 2016, and out-of-sample 12-month horizon forecasting from January 2017 to December 2017. The main research problem is that given the struggle policymakers have had to define proper criteria to diagnose the onset of inflation indicators, we felt compelled to identify an approach and methodology that the government of Montenegro can use in the threshold to accessing the European Union. We examine three individual-predictor SVAR models to forecast inflation.  Model 1 examines the internal determinants of inflation. Model 2 relates to demand-pull and cost-push variables. Model 3 examines external determinants. Combining the above three forecasts, we disclose two more RMSEs: equal and inverse MSE weights. Model 1 predicts inflation at 1.3%, the inverse MSE at 1.5%, and the weighted average at 1.4%. They show forecasting performances that are sustainable and average inflation not more than 1.5% above the rate of the three best performing Member states: Cyprus (0.2%), Ireland (0.3%), and Finland (0.8%) over the 12 months covering April 2017-March 2018. Our findings allow the policymakers to understand the factors involved in identifying the onset of inflation dynamics and inflation expectations in Montenegro better and develop more effective government regulations that can be employed nationally. In so doing, this research advances and recommends the toolset needed, combining forecasts, to combat the concerns of many macroprudential policymakers in Montenegro, especially the Central Bank of Montenegro.

Additional Files